To understand the role of frontal cortex in motor sequence learning we compared the effects of motor (M1), premotor (M2) and midline frontal (MFr) cortical lesions on rats making nose-pokes guided by luminance cues. Organizational demands were manipulated by varying the number (1 vs. 5) and predictability (random vs. repeated) of nose-pokes in a response. Learning was studied by comparing sessions with random or repeated cues. All cortical lesions increased reaction time (RT) during response initiation. These effects were larger for nose-pokes initiating sequential responses but spared RT for nose-pokes completing them. Repetition learning had significant effects on the speed and accuracy of single nose-poke responses that were unaffected by any of the cortical lesions. Repetition learning had more complex effects on sequential responding. RTs increased for nose-pokes initiating sequences over several sessions of continuous repetition and then decreased or leveled off. RTs decreased incrementally across all repetition sessions for subsequent nose-pokes in repeated sequences, following a time-course consistent with habit learning. Lesions involving M2 and MFr cortex exacerbated the increase in RT during initiation without affecting the incremental decrease in RT for nose-pokes completing repeated sequences. These results were confirmed by analyses of interference effects when training shifted from repeated (learned) to random (novel) sequences or to a new repeated sequence. These results implicate dorsomedial frontal cortex in organizational aspects of sensory-guided responding and motor sequence learning reflected in RT during response initiation.