Several self-matched approaches have been proposed, including case-crossover, case-time control, fixed-effects case-time control, and self-controlled case series. Rather than comparing treatment effects between different individuals, studies use these approaches to evaluate the acute effects of transient exposures, often called “triggers,” by comparing outcome risk among the same individual at different times. This eliminates confounding by between-person characteristics that remain stable over time, allowing for valid analyses even in situations where information on some health behaviors is not available, such as long-term smoking history. However, to attain valid results, differences in the probability of exposure and outcome that change over time must be addressed in the design and analysis of the study. In this article, we describe the setting, assumptions and analytic options for conducting studies using self-matched data. Approaches that involve matching or a group of noncases to address time-varying confounding may have less statistical flexibility but they are powerful tools that overcome the need to assume a particular form of any time trends in potential confounders. If data are available for all of the person–time under study, there is a gain in statistical efficiency and the ability to address time-varying confounding using flexible regression models, under the strong assumption of no mis-specification of the model.