Epidemiological studies analyze the occurrence and distribution of disease among populations. In general, these studies aim to determine the association (if any) between exposures to suspected disease-causing agents and health effects by comparing populations. There are three types of common epidemiological studies. Case-control studies compare exposures of people who have a certain disease against those who donŐt. Cohort studies examine the differences in disease rates between exposed and non-exposed populations. Ecological studies study the rate of disease of a population in a given geographical area based on average measures of exposure. Because ecological studies are not based on the actual exposure of individuals, they are less sophisticated than the other two types of studies, and results should be treated with caution.

In cohort studies, where a well-defined exposed population exists, epidemiologists calculate the relative risk or risk ratio of the exposed population by examining the rate of disease or death among exposed populations and dividing it by the rate in non-exposed populations. Epidemiological studies may also compare the number of cancer deaths in a studied population with the rates among the general population. The attributable or excess risk is calculated by taking the the difference (as opposed to ratio) between disease or death rates between exposed and non-exposed populations. In all epidemiological studies, it is important that the populations studied be adjusted for factors such as age, gender, and lifestyle habits (such as smoking), because disease rates can differ greatly across different groups.

Dose reconstruction studies estimate the exposure of individuals or of a population to a disease-producing agent like radiation. In order to estimate exposure, it is essential to know the amount of a pollutant released to a particular medium, such as air or water, from a source of pollution (called a source term), or to have an accurate history of concentrations of pollutants in air, water, and soil. Pathway analysis clarifies the often complex ways in which pollutants reach people through the environment, allowing release estimates to be converted to dose estimates. For example, pollutants can be simultaneously inhaled from the air, and ingested through drinking contaminated water or eating contaminated food. In addition, a population may receive both external and internal doses. Dose reconstruction studies can be conducted independently of epidemiological studies, but they can also help epidemiologists group together exposed populations more precisely.

Dose reconstruction and epidemiological studies can be powerful tools in determining the relationship between a pollutant and a health outcome. However, there are a number of complications which can cloud their results.

• Incorrect or incomplete data on pollutants. In general, it is easier to estimate doses to workers, who are often subject to some kind (if often inadequate) monitoring, than to neighboring populations, for whom data are generally not available. However, data on non-radioactive toxic materials are often lacking for workers, as well as for off-site populations.
• Difficulty in separating exposed from non-exposed populations. Populations may be incorrectly grouped because of poor or incomplete records. For example, worker groups have been grouped based on external doses from beta and gamma radiation, often because internal dose data are lacking. If the exposed people cannot be grouped into appropriate dose ranges, then estimation of increase in risk becomes very difficult. This is especially the case if a small proportion of highly exposed people are mixed in with a far larger number of people with relatively low exposure.
• Difficulty in tracking individuals over long periods. The period of time between exposure and the onset of disease (known as latency) can stretch over many decades, such as in the case of cancers or generational effects like birth defects. It is easy to lose track of individuals in these periods as they move, marry, change jobs, etc.
• Misdiagnosis and/or incorrect documented cause of death. Uncertainties posed by interaction between environmental or occupational exposure and other factors like gender, age, diet, smoking and other lifestyle habits.
• Synergistic effects. Populations are often exposed to more than one disease-producing agent, and the synergistic effects of these agents are not well-known.
• A disproportionate focus on cancer effects. Non-cancer effects of toxic materials, like birth defects and immune system damage, are only now beginning to be understood, and are therefore often overlooked.
• Small sizes of exposed populations combined with the low background occurrence of many diseases lead to large statistical uncertainties. Since there are considerable differences in they way different people respond to disease-producing agents, there must be sufficient numbers of people in an epidemiological study to determine with a reasonable certainty if there is an increased risk.
• For a variety of reasons, there are usually large uncertainties about the health effects of low levels of exposure to radiation and other toxic materials.