Biological plausibility is the idea that a proposed cause-and-effect relationship in health makes sense given what we currently know about how the body works. If a study finds that a certain chemical exposure is linked to cancer, biological plausibility asks: is there a known mechanism, a credible chain of events inside cells or tissues, that could explain how that chemical actually causes cancer? It’s one of nine criteria proposed by the epidemiologist Sir Austin Bradford Hill in 1965 to help scientists decide whether a statistical association between an exposure and a disease reflects a true causal relationship or is just a coincidence.
How It Fits Into Causal Reasoning
Finding a correlation in health data is relatively easy. People who live near highways have higher rates of asthma. People who drink more coffee report fewer cases of a particular disease. But correlation alone doesn’t tell you whether one thing actually causes the other. Bradford Hill laid out a set of considerations to help researchers weigh the evidence, including the strength of the association, whether it’s been seen consistently across studies, whether higher doses produce bigger effects, and whether the timeline makes sense (exposure before disease, not after).
Biological plausibility sits alongside these other considerations as a kind of reality check. It asks whether the relationship is consistent with the current body of knowledge about how diseases develop and how the body responds to exposures. Hill himself acknowledged a key limitation: plausibility depends entirely on the state of scientific knowledge at the time. A relationship might look implausible simply because researchers haven’t yet discovered the mechanism behind it.
What Counts as Evidence of Plausibility
Plausibility is judged based on whether existing biological or social models can explain the association in question. Before the tools of molecular biology became widespread, researchers often treated the connection between an exposure and a disease as a “black box.” They could see that people exposed to something got sick more often, but the biological steps in between were unknown and left out of study designs entirely.
Modern molecular epidemiology has changed this. Researchers can now trace the steps between exposure and disease at the cellular level, using lab studies in human cells, animal models, and measurements of biological markers in people. For example, when scientists investigated whether benzene causes a type of blood cancer called acute myeloid leukemia, they identified the specific metabolic byproducts of benzene that damage cells. They showed that one of these byproducts triggers cellular changes consistent with the early stages of that cancer in humans. This kind of molecular-level evidence filled in the black box and strengthened the case for causation without needing yet another observational study.
Similarly, when epidemiological studies linked exposure to polychlorinated biphenyls (PCBs) to melanoma, lab studies in human skin cells showed a plausible mechanism: PCBs disrupt the process by which skin cells produce pigment. The consistency between the population-level data and the cellular-level findings supported a causal interpretation.
Plausibility vs. Coherence
Hill’s criteria include a separate consideration called “coherence,” and the two are often confused. The distinction is subtle but meaningful. Plausibility asks: can you imagine a mechanism that, if it were operating, would produce the results you’re seeing in the data? Coherence asks something slightly different: do the results fit into the established theory without contradicting it?
In practice, coherence is more conservative. It would reject a finding as non-causal if it contradicted a dominant theory. Plausibility gives researchers more room because it doesn’t require the finding to align with one specific theory. It only requires that some credible biological explanation could exist.
The Smoking and Lung Cancer Example
The relationship between cigarette smoking and lung cancer is the classic case where biological plausibility played a decisive role. Early epidemiological studies in the 1950s showed a strong statistical link, but skeptics demanded a mechanism. Over the following decades, researchers identified it in precise detail.
Cigarette smoke contains carcinogens that the body tries to break down using its normal detoxification enzymes. But during that breakdown process, reactive chemical intermediates are produced. These intermediates bind to DNA, forming what are called DNA adducts. If the cell’s repair machinery doesn’t fix the damage before the DNA is copied, the wrong genetic “letter” gets inserted during replication, creating a permanent mutation. When that mutation lands in a gene that controls cell growth, such as the tumor suppressor gene TP53 or the oncogene KRAS, the cell loses its ability to regulate its own division. The result is cancer.
This chain of events, from chemical exposure to enzyme activation to DNA damage to mutation to uncontrolled growth, provided textbook biological plausibility. It transformed a strong statistical association into a widely accepted causal conclusion.
When Plausibility Gets It Wrong
The biggest limitation of biological plausibility is that it can only reflect what scientists know right now. If the mechanism behind a real causal relationship hasn’t been discovered yet, the relationship will look implausible, and legitimate findings may be dismissed.
The story of the bacterium that causes stomach ulcers is the most famous example. For most of the 20th century, the stomach was assumed to be a sterile environment where bacteria could not survive because of its extreme acidity. When Barry Marshall and Robin Warren proposed in the 1980s that a bacterium (now called Helicobacter pylori) was responsible for most stomach ulcers, the idea was considered biologically implausible. The prevailing model said ulcers were caused by stress and excess acid. It took years of additional evidence, including Marshall’s dramatic self-experimentation, before the medical community accepted the bacterial cause. The old model of plausibility had been wrong because the underlying biology wasn’t yet understood.
This is why Hill himself warned against placing too much weight on plausibility alone. As one recent analysis put it, it would be premature for any reasonable scientist to dismiss a causal relationship on the grounds of inadequate evidence for biological plausibility.
The Electromagnetic Fields Debate
A more recent example shows how plausibility arguments can be misused. When epidemiological studies in the 1980s and 1990s suggested a possible link between electromagnetic fields from power lines and cancer, physicists and engineers objected that the findings “must be wrong” because they contradicted well-established principles of biophysics. The energy levels involved were too low to damage DNA through any known mechanism.
But this reasoning contains a logical error. If those epidemiological studies found a spurious result, the problem had to lie within the studies themselves, perhaps selection bias or measurement error. The source of the error cannot be the contradictory evidence from other fields. Mechanistic evidence and epidemiological evidence are independent contributors to a causal assessment. One doesn’t invalidate the other; they are weighed together. The electromagnetic fields case remains debated, but it illustrates why plausibility should inform the overall picture rather than serve as a veto.
How Dose-Response Strengthens the Case
One of the considerations that works hand-in-hand with biological plausibility is the dose-response relationship, sometimes called the biological gradient. If more exposure leads to more disease in a predictable way, it becomes harder to explain the association as a fluke. Hill pointed to the fact that lung cancer death rates rise linearly with the number of cigarettes smoked daily as something that “adds a very great deal” to the simpler observation that smokers die more often than non-smokers.
The GRADE system, a widely used framework for rating the certainty of evidence in medicine, upgrades its confidence in a causal relationship when a dose-response pattern is present. The reasoning is that confounding (some hidden third factor causing both the exposure and the disease) is less likely to produce a clean dose-response curve. That said, dose-response alone isn’t proof. If confounding variables happen to track closely with the exposure, they could mimic a gradient.
Its Role in Drug Approval
Biological plausibility isn’t just an academic concept. It has practical consequences in regulatory decisions. The U.S. Food and Drug Administration uses what it calls a “plausible mechanism framework” when evaluating certain individualized therapies, particularly those designed to target specific genetic conditions with a known biological cause. Under this framework, developers must show that their drug’s mechanism of action makes biological sense for the condition being treated, supported by both lab studies and clinical data. This is especially relevant for rare genetic diseases where large clinical trials may not be feasible, and regulators need to weigh mechanistic evidence more heavily than they normally would.
Why It Still Matters
Sixty years after Bradford Hill introduced his criteria, biological plausibility remains central to how scientists, regulators, and public health officials evaluate health risks. It has also become harder to apply well. Science is increasingly specialized, and the experts who study cellular mechanisms often work in entirely different fields from the epidemiologists who study population-level patterns. Integrating these different lines of evidence requires deliberate effort across disciplines.
The core idea, though, is simple: when you find a statistical link between something in the environment and a health outcome, you should be able to tell a credible biological story about how it could happen. That story doesn’t need to be proven in every detail, but it should be grounded in what we know about how cells, organs, and biological systems work. And when the story doesn’t exist yet, that’s a reason for more research, not a reason to dismiss the findings.