Four major epidemiological study designs are reviewed: cohort, case-control, case-cohort, and case-crossover. In the medical field, these study designs and their analysis methods are commonly used to estimate the effect of exposure to a disease on an outcome (such as death). The formal epidemiological definition of each design in the medical field is here translated into the context of real-world and naturalistic driving safety studies. For example, instead of an outcome of death, the outcome becomes a crash or other safety-relevant event. Instead of exposure to a disease, the exposure becomes a driver activity such as a secondary task (e.g., talking on a cell phone), a driver impairment (e.g., drunk or drugged), or a driver behavior error (e.g., speeding). The effect size measures of the exposure on the outcome include the rate ratio, risk ratio, and odds ratio. Careful selection of a study design and the appropriate analysis method for that study design is vital to obtaining a valid effect size measure. It is also critical to control or adjust for factors that can bias the effect size, including driver demographic factors (e.g., sex, age, driving experience), and environmental factors (e.g., traffic density, closeness to junction, weather). An incorrect design or analysis method, or insufficient adjustment for biases, will give rise to an invalid estimate of the population effect size. Examples of applying the four epidemiological study designs to driving safety will be presented using hypothetical data, naturalistic driving study data, and real-world study data.