Epidemiologists recognize that systematic errors in the design or conduct of a study may bias the results. Information on the exposure of interest may be especially prone to misclassification. Even information that has been well-documented may be reported incorrectly. Study subjects may have difficulty recalling past exposures or behaviors, or may provide responses based on wishful thinking. The nature and importance of these biases is not always fully considered by investigators in their data analysis and in their investigation. This paper reviews the most common type of biases and cites examples of how the responses of subjects substantially affect study results.