In many fields of science, phenomena are studied closely over time to make inferences about patterns of behavior and to allow for predictions of future change and stability. In behavioral medicine, traditional cross-sectional and longitudinal designs are useful for capturing highly stable or slowly-changing phenomena, but important behavior change can be missed by one-occasion measures or infrequent measurements taken at widely-spaced points in time. We review recent research showing how a more complete understanding of many forms of psychological and somatic phenomena can be achieved through intensive measurement within the temporal context in which behavior occurs. This “process approach” to research, conducted in both naturalistic and laboratory settings, is presented here and placed in a methodological and theoretical framework. We also attempt to make recent research on the non-linear dynamics of behavior more accessible by describing and illustrating the uses of time in behavioral medicine research.