The type of representations involved in implicit learning during the response preparation stage of sequence learning was studied using the response-locked lateralized readiness potential (LRP-R) as an index. Participants performed a modified serial reaction time task on an eight-letter sequence; half were informed to look for patterns in the sequence and half were not. The standard sequence was occasionally replaced by one of two deviant sequences and the LRP-R elicited was analyzed. When comparing across three sequence conditions, namely, a perceptual deviant sequence, a motor deviant sequence, and a standard sequence (as control), the onset latency and the mean amplitude of the LRP-R were similar for all three conditions and for both explicit and implicit learners. The perceptual deviant sequence group showed a longer response preparation time than the motor deviant group and the standard sequence group, but the latter two were not significantly different. The perceptual deviant sequence group showed more negative LRP-R amplitude than the other two groups that did not differ from each other. The finding of similar LRP-R profiles between implicit and explicit learners suggests that stimulus representation (S–S association) is the main form of representation supporting sequence learning in the response preparation stage.