Habits control much of our waking behavior and are generally composed of sequences of actions of which we are hardly aware. Multiple basal ganglia circuits are thought to function together during the learning and performance of such behaviors, with the dorsolateral striatum being particularly critical for setting up and maintaining habitual behaviors. However, how neural activity in the dorsolateral striatum subserves these functions is poorly understood. To explore the nature of striatal neural representations during the performance of acquired action sequences, I designed a sequential lever press task specifically targeted at disambiguating lever press movement-related responses from habit-related representations in the spike activity of neurons in the dorsolateral striatum. In simultaneous motor cortical and striatal recordings, I found that, unlike neurons in the motor cortex, which tend to represent individual movements, neurons in the dorsolateral striatum preferentially encoded the start and end of the learned action sequence within which individual actions occurred. This striatal activity pattern generalized across a wide variety of movement sequences learned by different rats. Remarkably, when rats spontaneously performed unreinforced sequences of lever presses, the same neurons failed to exhibit the task-boundary activity. The results of optogenetic experiments and neural recordings suggested that motor cortex was not a primary driver of the striatal start-and-end activation, further drawing a distinction between task-related neural activity in striatum and motor cortex. Instead, I found that local striatal inhibitory interneurons are likely involved in shaping the task-boundary activity by increasing their firing rates mid-trial when projection neuron activity was suppressed between the start and end of the learned sequences. In comparison with the dorsolateral striatum and motor cortex, regions in the medial corticostriatal circuit, including dorsomedial striatum and prelimbic cortex, were weakly activated by the lever press sequence task, further highlighting the importance of the dorsolateral basal ganglia circuit in the performance of stereotyped action sequences. These experiments provide a definitive test for the existence of a striatal signal that could underlie motor chunking, selectively marking behavioral units consisting of learned action sequences. This activity could enable the learning and expression of habitual action sequences, a prime function of basal ganglia circuits.