5G Scientific Evidence: Research Gaps in 2026
5G Scientific Evidence: Bodewein and colleagues examine the physiological and health-related effects of 5G-relevant radiofrequency fields in humans and mammals in a systematic review published in PLOS ONE on 4 November 2026. The review is titled Physiological and health-related effects of 5G-relevant radiofrequency fields: A systematic review of experimental studies in humans and mammals.[1]
The review provides a useful overview of the available experimental research, while highlighting important gaps concerning study quality, realistic exposure conditions and the absence of qualifying epidemiological studies.
An important distinction: The review identified 107 studies that met its original eligibility criteria. The researchers later identified 28 additional articles in a supplementary search conducted in March 2026. They did not assess these additional articles in the same detail or include them in the main evidence synthesis.
Where does the evidence stand?
The main findings are:
- Study quality remains a concern. The researchers assigned 57 of the 107 studies to the lowest-quality category and included the remaining 50 in their main evidence assessment.
- Research specifically investigating 5G remains limited. Only five of the 50 studies included in the main evidence synthesis explicitly stated that they were investigating 5G exposure.
- There is a lack of epidemiological evidence. No epidemiological studies met the review’s eligibility criteria. This means the review could not assess possible health effects in human populations over time.
- The researchers have not fully assessed the additional studies. The 28 articles identified in the March 2026 supplementary search were not included in the main evidence synthesis.[1]
- The newer human studies provide additional observations. The five studies identified in the March search and the four papers published afterwards add to the experimental evidence, but do not resolve questions about long-term health effects.
- The newer findings do not represent four independent replications. Several of the four later papers arose from the same GOLIAT research programme, a relationship discussed in more detail below.
The important question is not simply whether individual studies report biological effects. It is whether the research as a whole is sufficiently reliable, independent and representative of real-world exposure to help assess possible health effects.
The evidence remains incomplete. Researchers need to conduct more rigorous studies of longer-term exposure and real-world 5G signals.
What did the researchers actually find?
The review focused on radiofrequency exposures in the 3–30 GHz range, rather than the full range of frequencies used by 5G networks.
The researchers considered 107 experimental studies involving humans and mammals:
- 100 animal studies;
- Seven controlled human studies.
The studies investigated 23 different topics, including cognition, behaviour, male fertility, immune responses, thermoregulation, skin and eye effects.
However, only 50 of the 107 studies were included in the main evidence synthesis. The remaining 57 were assigned to the lowest-quality tier because of their risk-of-bias assessments.
The review found that the available evidence was limited by methodological weaknesses, inconsistent findings and a shortage of experiments using exposure conditions representative of actual 5G technology.
The findings provide a useful assessment of the experimental evidence, but do not establish whether long-term exposure to 5G signals is safe or harmful to human health.[1]
Why is this a concern?
The gap in epidemiological research
One of the most important findings is the absence of epidemiological studies that met the review’s inclusion criteria.
Epidemiological research examines health outcomes in human populations. It can help investigate whether patterns of exposure are associated with particular health outcomes over time.
The absence of qualifying epidemiological studies does not mean that no epidemiological research on radiofrequency exposure exists. It means that the reviewers did not identify studies that met their specific criteria for this assessment.
Some studies were excluded because they examined mobile communications generally without separately assessing 5G exposure. Cross-sectional studies were also excluded because they cannot establish the temporal relationship between exposure and an outcome. Ecological studies without individual exposure assessments were not eligible either.[1]
The authors further note that there are almost no new or ongoing epidemiological projects specifically investigating the health effects of mobile-phone exposure.[1]
Why this gap matters
This leaves an important gap in the evidence. Experimental studies can investigate biological responses under controlled conditions, but they cannot replace long-term research in human populations.
Such research is important for examining possible health outcomes associated with repeated or prolonged exposure under real-world conditions.
How did the researchers assess the studies?
The review used the Office of Health Assessment and Translation (OHAT) risk-of-bias rating tool to assess the methodological quality of the studies.[1]
The researchers divided the 107 studies into three tiers:
- First tier – 11 studies (10%): Low risk of bias.
- Second tier – 39 studies (37%): Moderate risk of bias.
- Third tier – 57 studies (53%): High risk of bias.[1]
Only the first- and second-tier studies were included in the main narrative evidence synthesis. The third-tier studies were not simply ignored: their information is available in the supplementary material, but the authors did not consider them sufficiently reliable for the main assessment.
The risk-of-bias assessment identified several recurring weaknesses. These included inadequate reporting of allocation concealment and blinding, as well as insufficient descriptions of sham exposure conditions.
Blinding is important because researchers and participants should not know whether a real or sham exposure is being administered when that knowledge could influence the conduct or assessment of an experiment.
The authors found that many studies did not adequately describe the measures used to prevent such bias. This does not automatically mean that the results are incorrect, but it reduces confidence in the findings.
An interesting observation was that the proportion of studies reporting biological effects increased as methodological quality decreased:
- 59% of first-tier studies reported biological effects.
- 76% of second-tier studies reported biological effects.
- 90% of third-tier studies reported biological effects.[1]
The authors suggest that some findings in lower-quality studies may have resulted from factors that were not adequately controlled. This is an important possibility, but it does not demonstrate that all reported effects in those studies were false.
The distinction between a biological effect and an adverse health effect is also important. A measurable biological response does not necessarily mean that harm has occurred.
What did the higher-quality studies find?
The review identified only a small number of findings for which the authors considered the evidence sufficiently consistent to warrant discussion.[1]
Cytokine secretion in mice
The authors assessed the evidence for changes in cytokine secretion in mice as moderate.
Cytokines are signalling proteins involved in regulating immune and inflammatory responses. Some studies reported changes following exposure to radiofrequency fields below the applicable ICNIRP exposure limits.
However, the four studies contributing to this finding came from the same research group. Independent replication is therefore needed before firm conclusions can be drawn.
Structural changes in the rat hippocampus
The authors assessed the evidence for structural changes in the hippocampus of rats as low.
The hippocampus is a brain region involved in learning and memory. Some studies reported changes following exposure above ICNIRP limits, but the findings were inconsistent.
Neither of these findings establishes that 5G exposure causes adverse health effects in humans.
The thermal effect and higher frequencies
The review discusses the established thermal effects of radiofrequency exposure.
The review states that, to date, the only scientifically accepted biological effect of RF-EMF at frequencies above 10 MHz is an increase in temperature caused by the absorption of electromagnetic energy by body tissues. This thermal mechanism forms the basis of current exposure limits recommended by organisations such as the International Commission on Non-Ionizing Radiation Protection (ICNIRP) and the Institute of Electrical and Electronics Engineers (IEEE).
The authors also acknowledge that there is scientific and public debate about whether non-thermal effects may occur below these exposure limits, particularly following long-term exposure.[1]
However, the review found that temperature-related effects were rarely investigated systematically in the experimental studies.
Only five studies addressed thermoregulation or heat-related effects. Two rat studies, using frequencies of 26.5 and 28 GHz, reported effects at exposure levels above the applicable limits. Both came from the same research group, so independent confirmation is needed.[1]
Higher radiofrequencies are absorbed more superficially than lower frequencies. Millimetre-wave energy is largely absorbed in the skin, meaning that deeper organs receive much less direct exposure.
This makes research into possible effects on the skin and eyes particularly relevant as higher-frequency 5G systems are introduced.
It also highlights the importance of investigating actual exposure conditions rather than assuming that findings at one frequency can automatically be applied to another.
Do the studies reflect actual 5G exposure?
One of the most important limitations identified by the review concerns the relevance of the exposure signals used in the experiments.
Among the 50 first- and second-tier studies included in the main evidence synthesis, only five explicitly stated that they investigated 5G.[1]
However, the authors noted that even these five studies did not use typical 5G modulation patterns. Their signals were either continuous wave or GSM-like pulse-modulated signals. The remaining studies investigated therapeutic applications, high-power microwave exposures, radar signals or microwave exposures without focusing on a specific application.
This matters because 5G is not defined by frequency alone.
Signal characteristics, modulation, exposure duration and the way signals are transmitted all influence the exposure conditions. The review also notes that biological effects may depend on technical features such as beamforming and massive multiple-input multiple-output (MIMO) systems.
Consequently, findings from experiments using older or substantially different signals cannot automatically be treated as representative of exposure to modern 5G networks.[1]
What did the March 2026 search find?
The main evidence assessment was based on literature searches conducted in November 2022 and August 2023. To identify research published since then, the authors conducted an additional search of the EMF-Portal database on 18 March 2026.[1]
This supplementary search identified 28 relevant articles: five human studies and 23 animal studies.[1]
These articles were not subjected to the same detailed data extraction and risk-of-bias assessment as the original studies. They were therefore not incorporated into the main evidence synthesis.[1,2]
Of these 28 studies:
- 19 investigated frequencies at or below 6 GHz.
- Five investigated frequencies between 24 and 28 GHz.
- One covered both frequency ranges.
- Three investigated frequencies between 9.3 and 10 GHz.[1]
Seventeen of the 28 studies explicitly referred to 5G, including all five human studies and 12 animal studies.[1]
The 12 animal studies covered a range of topics, including brain and behavioural outcomes, male reproductive effects, offspring, eye effects, skin responses and stress biomarkers.[1]
The supplementary material provides a reference list of these newly identified articles, allowing readers to examine the additional research themselves.[2]
What about the five additional human studies?
The five human studies identified in the March 2026 supplementary search are worth examining individually because they provide some of the more recent experimental evidence involving 5G-relevant signals.
Importantly, these studies were identified after the original evidence assessment and were not included in the review’s detailed risk-of-bias assessment.
1. Jamal et al. (2024): Autonomous nervous system responses to environmental-level exposure to 5G’s first deployed band (3.5 GHz) in healthy human volunteers
This study involved 44 healthy young volunteers and investigated skin temperature and electrodermal activity during and after exposure to 3.5 GHz signals.
The researchers used a randomised, crossover, triple-blind design, comparing real and sham exposure sessions. The electric field strength was 2 V/m at head level and 1.5 V/m at trunk level.
The study found small changes in head and neck temperature and in certain electrodermal measurements. The authors noted that these changes remained within normal physiological ranges and could have been influenced by an uncontrolled factor.
The findings are therefore preliminary and require further investigation.[3]
2. Michelant et al. (2026): Brain electrical activity at 26 GHz
This randomised, triple-blind, crossover study involved 31 healthy young adults. Participants were exposed to 26 GHz radiofrequency fields for 26.5 minutes at an electric field strength of 2 V/m.
The researchers examined brain electrical activity using electroencephalography (EEG).
No significant changes were observed in the EEG measurements across the frequency bands or cortical regions investigated.[4]
3. Michelant et al. (2026): Salivary stress biomarkers at 26 GHz
A separate study by Michelant and colleagues examined whether short-term exposure to 26 GHz affected salivary cortisol and alpha-amylase, two biomarkers associated with stress and autonomic nervous system activity.
The study included 31 participants who underwent a triple-blind, randomised protocol involving 26.5 minutes of real or sham exposure. A subset of 16 participants also took part in an exploratory protocol involving cognitive testing and more frequent saliva sampling.
The researchers found no significant differences in salivary cortisol or alpha-amylase between real and sham exposure conditions.[5]
Both Michelant studies investigated short-term exposure to 26 GHz fields in groups of 31 participants. The publications describe closely related experimental protocols, so their findings should not be interpreted as evidence from two independent groups of 31 participants.
4. Sousouri et al. (2025): Sleep EEG and genetic variation
This study involved 34 participants and investigated whether a particular genetic variant, CACNA1C rs7304986, influenced the effects of 5G radiofrequency exposure on sleep EEG activity.
Participants underwent three experimental conditions: 700 MHz exposure, 3.6 GHz exposure and sham exposure. Each exposure lasted 30 minutes before sleep.
The researchers reported a genotype-dependent change in sleep spindle centre frequency following 3.6 GHz exposure. The increase was observed in participants carrying the T/C variant, but not in the T/T group.
Sleep spindles are patterns of brain activity that occur during non-rapid eye movement (NREM) sleep. The finding is a specific change in an EEG measure; it does not establish that exposure caused a sleep disorder or other adverse health effect.
The result warrants further investigation and replication.[6]
5. Torkan et al. (2025): Corticospinal and intracortical excitability
This randomised controlled pilot study involved 19 healthy adults. It investigated whether 3.6 GHz 5G mobile-phone exposure affected corticospinal excitability or intracortical mechanisms.
Participants underwent exposure conditions lasting five and 20 minutes, as well as sham exposure. Measurements were taken before and after exposure using transcranial magnetic stimulation.
The researchers found no statistically significant interaction between exposure condition and time for the measures investigated.
However, the Bayesian analyses produced Bayes factors close to 1, indicating inconclusive evidence for either the presence or absence of an effect.
It would therefore be inaccurate to interpret this study as demonstrating that 5G exposure has no effect. The results were inconclusive, and the authors recommended further research.[7]
How should these five studies be interpreted?
The five studies do not provide a consistent picture of biological responses. Some reported small physiological changes or a genotype-dependent EEG finding, while others found no statistically significant changes in the outcomes examined.
Three of the five human studies identified in the March 2026 supplementary search – Jamal et al. [3] and the two Michelant studies [4,5] – were associated with the GOLIAT research programme. Their findings therefore need independent confirmation.
These were controlled, short-term experiments involving healthy volunteers. They were not designed to investigate long-term health outcomes or determine whether the findings apply to the wider population.
Research published after the March 2026 search
Four additional human studies published during 2026 provide further information on physiological responses to 5G-relevant radiofrequency exposure. These studies appeared after the review’s supplementary literature search of 18 March 2026 and were therefore not included in its assessment.
The studies investigated different biological endpoints following short-term exposure to 3.5 GHz and 26 GHz signals under controlled laboratory conditions.
3.5 GHz exposure: Heart-rate variability and stress biomarkers
Jamal et al. (2026) investigated heart rate, heart-rate variability (HRV) and three salivary stress biomarkers: cortisol, alpha-amylase and chromogranin A. The study involved healthy young adults exposed to a 3.5 GHz signal.
The researchers found no consistent exposure-related changes in HRV or the salivary biomarkers. One statistically significant interaction involving the HRV parameter RMSSD was observed, but it was small, limited to a single measurement period and not supported by other indicators. The authors therefore cautioned against interpreting it as evidence of a consistent physiological response.[8]
26 GHz exposure: Skin temperature
Michelant et al. (2026) examined whether short-term exposure to 26 GHz radiofrequency signals affected skin temperature in 31 healthy young adults. Using infrared thermography, the researchers monitored temperature at several body sites during real and sham exposure sessions.
No statistically significant differences in skin temperature were attributable to RF exposure. Temperature decreases observed at some sites occurred independently of exposure condition.[9]
26 GHz exposure: Heart-rate variability
A second study by Michelant et al. (2026) investigated HRV responses in 31 healthy young adults exposed to 26 GHz signals.
The main HRV measurements showed no statistically significant differences between real and sham exposure. Although an interaction was observed for one parameter, low-frequency power, subsequent analyses did not identify significant differences between exposure conditions or experimental phases. The researchers considered this finding inconclusive.[10]
26 GHz exposure: Electrodermal activity
Michelant et al. (2026) also investigated electrodermal activity (EDA), which provides a measure of sympathetic nervous system activity, in healthy young adults.
The researchers found no consistent exposure-related changes in electrodermal responses. An isolated difference in response latency was already present before exposure and did not vary with the experimental phase, making an association with 26 GHz exposure unlikely.[11]
How should these additional findings be interpreted?
Taken together, the four papers did not demonstrate consistent acute physiological changes under the exposure conditions investigated. However, they add to the experimental evidence on biological responses to 5G-relevant frequencies and address endpoints that are not necessarily captured by conventional assessments of adverse health outcomes.
All four studies arose from the GOLIAT research programme, which forms part of the European CLUE-H cluster.[8–12] Although they investigated different physiological endpoints, they should not be regarded as four independent replications. The three studies on heart-rate variability, salivary biomarkers and electrodermal activity came from the same controlled human exposure research programme.
The studies involved healthy young adults and relatively short exposure periods. They therefore do not establish whether longer-term, repeated or cumulative exposure produces different responses, nor do they resolve questions about other populations or possible health outcomes.
These findings should be considered alongside the wider body of experimental research rather than as a definitive assessment of the biological or health effects of 5G exposure.
Limitations of the review
The authors acknowledge several limitations in their assessment.[1]
The review was restricted to peer-reviewed studies published in English or German. Grey literature and studies published in other languages were not included.
The review considered experimental studies involving humans and mammals, as well as epidemiological studies. In vitro studies, research involving non-mammalian species and studies deliberately examining combined exposure to RF-EMF and non-EMF factors were excluded.
The application of ICNIRP exposure categories to animal studies also involves uncertainty. These limits are developed primarily with human exposure conditions in mind and cannot always be transferred directly to other species.
What should happen next?
The review identifies several areas where further research is needed.[1]
Important areas for further investigation include:
- Better-designed, high-quality experimental studies.
- Adequate exposure characterisation and dosimetry.
- Research into possible effects on the skin and eyes.
- Studies using realistic 5G exposure scenarios and different field strengths.
- Epidemiological studies that consider 5G separately, where possible, preferably cohort studies using improved exposure assessment based on objective, prospective data.
The distinction between biological effects and adverse health effects must remain central to future research. The absence of qualifying epidemiological evidence should not be interpreted as proof that long-term effects do not exist.
Other research initiatives
Several European research projects have investigated questions relevant to 5G and radiofrequency exposure. ETAIN, GOLIAT, NextGEM and SEAWave are four projects within the European research cluster on electromagnetic fields and health, known as CLUE-H.[12]
The projects address different aspects of radiofrequency exposure and health, including exposure assessment and possible effects on the skin.[13] They have different timelines. ETAIN and GOLIAT are both scheduled to run until 31 May 2027.[14,15] NextGEM’s scheduled project period ended on 30 June 2026,[16] while SEAWave’s ended on 30 November 2025.[17]
These projects contribute to the broader research effort on electromagnetic fields and health. However, their work should not be confused with long-term epidemiological studies that examine health outcomes in populations over time.
ANSES: A new research initiative in France
There are also signs of renewed research activity at national level. In 2026, the French Agency for Food, Environmental and Occupational Health and Safety (ANSES) announced a new research call entitled Radiofrequencies and Health under its National Research Programme for Environmental and Occupational Health.[18]
The call covers frequencies from 8.3 kHz to 300 GHz and invites research into new exposure sources and signals, including connected devices, 5G, satellite technologies and different mobile communication modulations. It also allows projects to investigate radiofrequency fields in combination with other factors. The call places particular emphasis on methodological quality and research that can contribute to future expert assessments and public policy.
The deadline for submitting letters of intent was 1 October 2026. Researchers whose proposals are selected at this first stage must submit their complete applications by 11 March 2027, at noon French time. This is an encouraging initiative, but it is a call for research proposals, not evidence that new studies have already begun. Nor does it guarantee that epidemiological studies will be funded. It does, however, recognise the need to address outstanding questions about radiofrequency exposure and health.
References
Main review and supplementary material
[1] Bodewein L, Dechent D, Emonds T, Graefrath D, Keller AS, Krabbe J, et al. (2026). Physiological and health-related effects of 5G-relevant radiofrequency fields: A systematic review of experimental studies in humans and mammals. PLOS ONE, 21(11): e0361104. Published 4 November 2026.
https://doi.org/10.1371/journal.pone.0361104
[2] Bodewein et al. (2026). Supporting Information S12: Reference List of newly identified articles.
https://doi.org/10.1371/journal.pone.0361104.s012
Additional human studies identified in the March 2026 search
[3] Jamal L, Michelant L, Delanaud S, Hugueville L, Mazet P, Lévêque P, et al. (2024). Autonomous nervous system responses to environmental-level exposure to 5G’s first deployed band (3.5 GHz) in healthy human volunteers. Experimental Physiology, 109(12): 2122–2133.
https://doi.org/10.1113/EP092083
[4] Michelant L, Baz T, Carrie A, Hugueville L, Lévêque P, Selmaoui B. (2026). Millimeter-wave high frequency 5G (26 GHz) electromagnetic fields do not modulate human brain electrical activity. Environmental Research, 289: 123349.
https://doi.org/10.1016/j.envres.2025.123349
[5] Michelant L, Hugueville L, Lévêque P, Selmaoui B. (2026). No measurable impact of acute 26 GHz 5G exposure on salivary stress markers in healthy adults. Environmental Research, 290: 123439.
https://doi.org/10.1016/j.envres.2025.123439
[6] Sousouri G, Eicher C, D’Angelo RM, Billecocq M, Fussinger T, Studler M, et al. (2025). 5G radio-frequency-electromagnetic-field effects on the human sleep electroencephalogram: A randomized controlled study in CACNA1C genotyped volunteers. NeuroImage, 317: 121340.
https://doi.org/10.1016/j.neuroimage.2025.121340
[7] Torkan A, et al. (2025). The Effect of 5G Mobile Phone Electromagnetic Exposure on Corticospinal and Intracortical Excitability in Healthy Adults: A Randomized Controlled Pilot Study. Brain Sciences, 15(11): 1134.
https://doi.org/10.3390/brainsci15111134
Additional research on 5G-relevant radiofrequency exposure (2026)
[8] Jamal L, Michelant L, Delanaud S, Bodin R, Hugueville L, Mazet P, Lévêque P, Baz T, Stephan-Blanchard E, Selmaoui B. (2026). Exposure to 5G Radiofrequency and Physiological Effects in Healthy Young Adults: Insights Into Heart Rate Variability and Salivary Stress Biomarkers. Bioelectromagnetics, 47(5): e70056.
https://doi.org/10.1002/bem.70056
[9] Michelant L, Delanaud S, Hugueville L, Lévêque P, Bach V, Tourneux P, Selmaoui B. (2026). No thermal skin effects at environmental 26 GHz field strengths relevant to 5G deployment. Scientific Reports, 16: 19461.
https://doi.org/10.1038/s41598-026-49616-7
[10] Michelant L, Jamal L, Stephan-Blanchard E, Baz T, Leveille Nizerolle B, Hugueville L, Lévêque P, Selmaoui B. (2026). Heart rate variability responses to 26 GHz millimeter-wave exposure in healthy young adults. Electromagnetic Biology and Medicine. Published online 11 September 2026.
https://doi.org/10.1080/15368378.2026.2730027
[11] Michelant L, Hugueville L, Lévêque P, Selmaoui B. (2026). Effects of 26 GHz radiofrequency exposure on electrodermal activity in healthy young adults. International Journal of Radiation Biology, 102(10): 1392–1401.
https://doi.org/10.1080/09553002.2026.2678295
Research funding and calls
[12] European Commission. Cluster on Electromagnetic Fields Exposure and Health (CLUE-H). Official description of the four Horizon Europe projects and their collaboration.
[13] European Cluster on Electromagnetic Fields and Health (CLUE-H). Scientific Strategy of the Cluster. Further information about the research programme, including work on exposure assessment and skin-related effects.
https://www.emf-health-cluster.eu/wp-content/uploads/2024/09/CLUE-H_ScientificStrategy.pdf
[14] European Commission. Exposure To electromAgnetic fIelds and plaNetary health (ETAIN). CORDIS, Horizon Europe, Grant Agreement 101057216. Project period: 1 June 2022 – 31 May 2027.
https://cordis.europa.eu/project/id/101057216
[15] European Commission. 5G expOsure, causaL effects, and rIsk perception through citizen engAgemenT (GOLIAT). CORDIS, Horizon Europe, Grant Agreement 101057262. Project period: 1 June 2022 – 31 May 2027.
https://cordis.europa.eu/project/id/101057262
[16] European Commission. Next Generation Integrated Sensing and Analytical System for Monitoring and Assessing Radiofrequency Electromagnetic Field Exposure and Health (NextGEM). CORDIS, Horizon Europe, Grant Agreement 101057527. Project period: 1 July 2022 – 30 June 2026.
https://cordis.europa.eu/project/id/101057527
[17] European Commission. Scientific-Based Exposure and Risk Assessment of Radiofrequency and mm-Wave Systems from children to elderly (5G and Beyond) (SEAWave). CORDIS, Horizon Europe, Grant Agreement 101057622. Project period: 1 June 2022 – 30 November 2025.
https://cordis.europa.eu/project/id/101057622
[18] ANSES (2026). National Research Programme for Environmental and Occupational Health: Radiofrequencies and Health 2026 Call for Proposals.
https://www.anses.fr/sites/default/files/GB-Texte_APREST_2026-RF.pdf
This article discusses findings from published research. It does not claim that the reported biological effects establish adverse health outcomes or that the absence of statistically significant findings proves the absence of risk.
EMFSA | 9 November 2026