This paper describes and analyzes movements of older men among labor force states using quarterly observations derived from the Retirement History Survey (RHS). A comparison of transition rates using these quarterly records with the more typical biannual records from the RHS and other sources reveals substantial undercounts in the biannual data, indicating that the prevalence of labor force movements at older ages has been underestimated previously. The quarterly data reveal a much sharper spike in the exit rate at exact age 65 than is found in data measured at less frequent intervals. Estimates are presented of the parameters of a reduced form model of labor force transitions. A novel feature of the model is the exploration of dynamic aspects of labor force behavior at older ages. The results show that labor force dynamics at older ages are important, including duration and spell occurrence dependence, and work experience effects. These effects are robust to nonparametric controls for unobserved heterogeneity. The estimates indicate that social security benefits have strong effects on the timing of labor force transitions at older ages, but that changes in social security benefit levels over time have not contributed much to the trend toward earlier labor force exit.