The IARC classification system is a scheme developed by the International Agency for Research on Cancer, a specialized agency within the World Health Organization, to evaluate whether a substance, exposure, or situation can cause cancer in humans. It sorts agents into one of four groups ranging from “carcinogenic to humans” (Group 1) to “not classifiable” (Group 3), based on the strength of evidence rather than the degree of danger. That distinction trips up nearly everyone who encounters the system for the first time, and misunderstanding it has fueled public confusion over everything from processed meat to artificial sweeteners.
What the Four Groups Actually Mean
The IARC system has four main categories. Each reflects how confident the evidence is that an agent can cause cancer, not how much cancer it causes or how likely you are to get cancer from it.
- Group 1: Carcinogenic to humans. There is sufficient evidence from human studies that the agent causes cancer. This group includes tobacco smoking, asbestos, outdoor air pollution, and processed meat. Being in Group 1 does not mean all these agents are equally dangerous; it means the evidence that each one can cause cancer is equally strong.
- Group 2A: Probably carcinogenic to humans. The evidence from human studies is limited, but there is sufficient evidence from animal experiments or strong mechanistic data. Red meat, night shift work, and glyphosate have been placed here.
- Group 2B: Possibly carcinogenic to humans. The human evidence is limited and the animal evidence is also not fully convincing. Radiofrequency electromagnetic fields from mobile phones and aspartame fall into this category.
- Group 3: Not classifiable as to its carcinogenicity to humans. The evidence is inadequate in both humans and animals. This does not mean the agent is safe; it means there is not enough information to say one way or the other.
There was historically a Group 4 (“probably not carcinogenic”), but only one agent, caprolactam, was ever placed there, and the category has essentially fallen out of use. For practical purposes, the system operates on a four-tier scale from confident yes to we don’t know.
Hazard Identification, Not Risk Assessment
The single most important thing to understand about IARC classifications is that they answer a narrow question: “Can this agent cause cancer under any circumstances?” They do not answer the question most people actually care about: “How likely am I to get cancer from this?” That second question is risk assessment, and IARC deliberately does not do it. Risk assessment requires factoring in dose, duration of exposure, how people actually encounter the substance, and individual susceptibility. IARC’s process skips all of that.
This is not a flaw in the system. It is a design choice. IARC’s evaluations are meant to flag hazards so that national regulators and public health agencies can then perform their own risk assessments using local exposure data and policy priorities. But when the public hears that processed meat is in the same group as tobacco, the natural reaction is alarm, because most people read the classification as a statement about risk level. One study examining how people interpret IARC announcements found that the press releases conveying the classification are often far less nuanced than the full scientific monographs, and the distinction between hazard identification and risk assessment gets lost in translation.
How an Agent Gets Evaluated
The IARC Monographs programme has been running since the early 1970s, and over more than a hundred volumes it has evaluated over a thousand agents. The process begins with a nominated agent being prioritized based on evidence suggesting possible carcinogenicity and the level of human exposure. A working group of independent scientists from various disciplines is then assembled. These experts review published studies spanning three streams of evidence: epidemiological data from human populations, experimental data from animal studies, and mechanistic evidence showing how the agent might cause cancer at a biological level.
The working group rates the strength of evidence in each stream and then integrates all three to arrive at an overall classification. A revised set of procedures published in 2019 formalized how these three bodies of evidence are weighed together, aiming for greater transparency in how the final group assignment is reached.1PubMed Central. The IARC Monographs: Updated Procedures for Modern and Transparent Evidence Synthesis in Cancer Hazard Identification The updated procedures also placed more emphasis on systematic review methods and clearer documentation of the reasoning behind each decision, partly in response to criticisms that earlier evaluations were not always easy to audit from the outside.
The Key Characteristics Framework
One of the more significant methodological shifts in recent years has been the adoption of a framework built around “key characteristics” of carcinogens. Rather than asking broadly whether an agent causes cancer, this approach looks at whether the agent shows specific biological properties known to be associated with carcinogenesis. These properties include things like whether the agent damages DNA, causes chronic inflammation, alters cell growth, suppresses immune function, or disrupts hormone signaling.
The framework has been applied across a wide range of exposures, from individual chemicals and metals to more complex situations like occupational and dietary exposures. According to researchers involved in the programme, this approach has improved how mechanistic data is organized and evaluated, reduced bias in the evaluation process, and expanded the knowledge base around how known and suspected carcinogens actually operate at the cellular level.2PubMed Central. IARC Workshop on the Key Characteristics of Carcinogens: Assessment of End Points for Evaluating Mechanistic Evidence of Carcinogenic Hazards Before this framework, mechanistic evidence was often evaluated in a less structured way, making it harder to compare across agents or ensure consistency between different working groups.
Processed Meat, Outdoor Air Pollution, and Other Group 1 Agents
Some of the most widely discussed IARC classifications involve agents that billions of people encounter routinely. In 2015, IARC classified processed meat as a Group 1 carcinogen and red meat as Group 2A, based on epidemiological data linking both to colorectal cancer and on mechanistic evidence from animal and human studies.3PubMed Central. Mechanistic Evidence for Red Meat and Processed Meat Intake and Cancer Risk: A Follow-up on the International Agency for Research on Cancer Evaluation of 2015 The public reaction was intense, with headlines implying that eating a ham sandwich was comparable to smoking. That comparison misses the point entirely. The Group 1 label means the evidence is strong that processed meat can cause colorectal cancer, not that it causes as many cancers as tobacco does. The actual magnitude of risk from processed meat is far smaller.
Outdoor air pollution and the particulate matter within it were classified as Group 1 in 2013, based on sufficient evidence from human and animal studies along with strong mechanistic support.4PubMed Central. The International Agency for Research on Cancer (IARC) evaluation of the carcinogenicity of outdoor air pollution: focus on China Unlike processed meat, this classification generated relatively little public controversy, even though air pollution is arguably harder for individuals to avoid. The difference in reaction likely reflects the fact that people feel they have a choice about what they eat but less control over what they breathe.
When IARC and Regulators Disagree
The classification of glyphosate, the world’s most widely used herbicide, as Group 2A in 2015 sparked one of the fiercest scientific and regulatory controversies in recent memory. IARC concluded that glyphosate is probably carcinogenic to humans, while the European Union’s own assessment and a joint WHO/FAO evaluation reached different conclusions. A detailed review of the scientific basis behind these disagreements found that part of the divergence stemmed from the use of different data sets, particularly regarding long-term animal studies, but that methodological differences in how the available evidence was evaluated also played a substantial role.5PubMed Central. Glyphosate toxicity and carcinogenicity: a review of the scientific basis of the European Union assessment and its differences with IARC
A similar split emerged with aspartame. In 2023, IARC classified the artificial sweetener as Group 2B (possibly carcinogenic), citing limited evidence from animal studies, while JECFA, the joint WHO/FAO food safety committee, simultaneously concluded that aspartame is safe within established daily intake limits. One analysis of this disagreement noted that the divergent conclusions reflect the fundamentally different questions each body is asking: IARC asks “can this cause cancer under any conditions?” while JECFA asks “does this cause cancer under real-world conditions of use?”6Oxford Academic (Toxicological Sciences). Is statistical re-evaluation of hemolymphoreticular neoplasms from aspartame studies valid? Both answers can be technically correct and still send completely different signals to the public.
Glyphosate was not the first time IARC and national regulators reached opposing conclusions, but the intense public and legal attention it attracted made the disconnect more visible than ever. Lawsuits citing the IARC classification resulted in billions of dollars in jury awards against the herbicide’s manufacturer, even as regulatory agencies in multiple countries continued to approve its use. The episode highlighted a real tension in how hazard-only classifications interact with legal and regulatory systems that ultimately need to make binary decisions about whether a product stays on the market.
Mobile Phones and the Group 2B Debate
In 2011, IARC classified radiofrequency electromagnetic fields associated with wireless phone use as Group 2B, meaning possibly carcinogenic to humans.7PubMed Central. Mobile phone use and possible cancer risk: Current perspectives in India The classification generated global headlines and considerable anxiety, even though Group 2B represents the weakest positive classification in the system. It says there is some limited evidence suggesting a possible link, not that one has been established.
Research published after the 2011 evaluation has added complexity. Some epidemiological studies, particularly case-control studies from Sweden and France, have reported an association between long-term mobile phone use and glioma, a type of brain cancer. Some researchers have argued that the accumulated evidence since 2011 warrants upgrading the classification to Group 1.8PubMed. Cancer epidemiology update, following the 2011 IARC evaluation of radiofrequency electromagnetic fields (Monograph 102) Others dispute this interpretation, pointing to large cohort studies and time-trend analyses that have not consistently shown an increase in brain cancer rates despite the massive rise in mobile phone use over the past two decades. IARC has indicated it plans to re-evaluate radiofrequency fields, and the outcome of that reassessment will be closely watched.
The mobile phone example illustrates a recurring pattern with Group 2B classifications: the label itself is scientifically cautious, but in public discourse it gets compressed to “phones might cause cancer,” which for many people is functionally indistinguishable from “phones cause cancer.” Research on how the public interprets IARC press releases has confirmed that the difference between hazard identification and actual risk estimation is poorly communicated, with the short public-facing announcements carrying far less nuance than the full monographs.
Night Shift Work as a Carcinogen
One of the more surprising entries on the IARC list is night shift work. In 2007, IARC classified shift work that involves circadian disruption as Group 2A, based on limited human evidence and sufficient animal evidence.9Occupational and Environmental Medicine. Considerations of circadian impact for defining ‘shift work’ in cancer studies: IARC Working Group Report In 2019, after further review, a new working group maintained the Group 2A classification for “night shift work” specifically, citing strong mechanistic evidence in experimental animals and limited evidence from human epidemiological studies.10PubMed Central. IARC 2019: “Night shift work” is probably carcinogenic: What about disturbed chronobiology in all walks of life?
The mechanisms involve disruption of circadian rhythms, suppression of melatonin, and downstream effects on immune function, DNA repair, and hormone regulation. What makes this classification unusual is that “night shift work” is not a chemical or a form of radiation; it is a pattern of behavior affecting hundreds of millions of workers worldwide. That breadth of exposure means the public health implications, if the link is confirmed in stronger human studies, could be enormous. It also raises difficult policy questions, because unlike removing a chemical from commerce, you cannot simply eliminate night shifts from sectors like healthcare, transportation, and emergency services.
Criticisms of the IARC Process
The IARC Monographs programme commands significant scientific respect, but it has drawn criticism on several fronts. One persistent critique concerns the asymmetry of outcomes: very few agents are ever downgraded or placed in Group 3, which to some observers suggests the system is better at flagging hazards than at clearing agents that do not appear to be carcinogenic. Critics have noted that the binary nature of carcinogenicity classification schemes more broadly, not just IARC’s, tends to categorize substances as either carcinogenic or not, without adequately capturing the nuances of dose, exposure duration, and species-specific responses.
A more pointed concern involves how working group members are selected. IARC chooses experts based on subject-matter expertise, which typically means prior publication on the substance being evaluated. One analysis argued that this practice creates a significant potential conflict of interest, because experts who have previously published on a substance may bring strong pre-existing views about its toxicity. The same analysis also raised concerns about the practice of restricting working groups to selected publications rather than making all accessible data available for review.11PubMed. Conflicts of interest in the International Agency for Research on Cancer process of identifying carcinogenic hazards to humans IARC has policies to manage financial conflicts of interest, but intellectual or academic conflicts receive less formal scrutiny.
Defenders of the programme counter that subject-matter expertise is precisely what makes evaluations credible, and that excluding researchers who have published on a substance would leave working groups with evaluators who lack the depth of knowledge needed to interpret complex data. The balance between expertise and independence is a genuine tension without an easy resolution.
Re-evaluations and Changing Classifications
IARC classifications are not permanent. Agents can be re-evaluated as new evidence emerges, and some have moved between groups. Coffee is one of the best-known examples. It was classified as Group 2B (possibly carcinogenic to the bladder) in 1991, but a 2016 re-evaluation found that the earlier bladder cancer link was likely confounded by smoking, and coffee was reclassified to Group 3. In the same 2016 evaluation, very hot beverages (above about 65°C) were classified as Group 2A, shifting the concern from the drink itself to the temperature at which it is consumed.
These re-evaluations underscore an important feature of the system: a classification reflects the state of the evidence at the time it is made. As study methods improve, as larger cohorts are followed, and as mechanistic tools become more refined, the evidence base shifts. Agents classified decades ago may be due for a fresh look. The planned re-evaluation of radiofrequency electromagnetic fields is one such example, and the incorporation of the key characteristics framework into the evaluation process means that future assessments will likely draw on a broader and more structured range of mechanistic data than earlier ones did.
Occupational Carcinogens and Global Inequality
Many of the agents on IARC’s list are primarily encountered in occupational settings, including asbestos, crystalline silica, benzene, and diesel engine exhaust. These classifications have had direct effects on workplace regulation in high-income countries, leading to bans, exposure limits, and protective equipment standards. But the burden of occupational cancer is not evenly distributed around the world. From 1990 to 2021, high-income regions experienced the most substantial declines in cancer burden from occupational carcinogen exposure, while low-to-middle-income regions saw the greatest increases.12PubMed Central. Global Burden, Projection, and Inequalities Analysis of Cancer Attributable to Occupational Carcinogen Exposure in Individuals Aged Over 40 Years
This pattern reflects a familiar dynamic in global health: wealthy countries identify a hazard, regulate it domestically, and sometimes export the hazardous industries or products to countries with weaker regulatory infrastructure. A separate analysis of the global burden of cancers linked to occupational factors found that regions like South Asia and sub-Saharan Africa showed the steepest increases in cancer burden from workplace exposures, while high-income countries like the Netherlands, the United Kingdom, and Italy carried the highest absolute rates of occupational cancer deaths in 2021.13PubMed Central. The global burden of cancers attributable to occupational factors, 1990-2021 The IARC classifications themselves are global, but their protective effects depend entirely on whether national governments translate those classifications into enforceable regulations, and the gap between classification and enforcement remains wide in much of the world.