This study presents a dynamic model of how animals learn to regulate their behavior under time-based reinforcement schedules. The model assumes a serial activation of behavioral states during the interreinforcement interval, an associative process linking the states with the operant response, and a rule mapping the activation of the states and their associative strength onto response rate or probability. The model fits data sets from fixed-interval schedules, the peak procedure, mixed fixed-interval schedules, and the bisection of temporal intervals. The major difficulties of the model came from experiments that suggest that under some conditions animals may time 2 intervals independently and simultaneously.