The contextual interference hypothesis proposes that when learning multiple skills, massing practice leads to better within-day acquisition, whereas random practice leads to better retention and transfer. In this experiment, we examined the effect of practice pattern on the learning, consolidation (retention), and transfer of visual-motor sequences. On Day 1, participants were randomly assigned to the Massed, Alternating, or Random condition. On Day 2, all participants were tested for consolidation and transfer. Learning was assessed through changes in accuracy and response synchronization. We found that massed practice led to enhanced sensorimotor integration and timing (as measured by response synchronization), whereas random practice led to better stimulus-response association (as measured by accuracy). On day 2, all groups showed consolidation for both measures, as well as transfer for accuracy but not response synchronization. Overall, this pattern of results provides limited support for the contextual interference hypothesis. Our findings are consistent with differential encoding of specific domains of motor performance. We propose that learning of the more explicit stimulus-response association is a fast process that benefits from random practice because it requires the acquisition of this association in multiple contexts. Once the association is learned, it seems resistant to interference and transferrable to a novel sequence. In contrast, learning of the sensorimotor integration and timing is a slower process that benefits from blocked training because practice in a single context allows fine-tuning of the response. Given that all groups showed consolidation, we postulate that learning that occurs in the context of interference can show consolidation.