Many studies have suggested that under some conditions sequence learning may be implicit and outside consciousness. Others suggest that a sequence can be learned with minimal involvement of attention when it contains unique associations between some elements. P. Perruchet and M. A. Amorim (1992) have recently questioned these suggestions and concluded on the basis of 3 experiments that consciousness is necessary for learning sequences and that unique associations may not be essential in learning with distraction. We outline several flaws in the arguments made by Perruchet and Amorim: They equated tasks with knowledge, assumed that correlations under some conditions preclude dissociations under many other conditions, and misinterpreted claims concerning unique associations in a sequence. We claim that the experiments and analyses conducted by Perruchet and Amorim are not diagnostic to these issues, so we view their conclusions as unwarranted.