Memory for object motion in 3-month-old infants was investigated across retention intervals of 1 or 3 months in three studies using a novelty preference method. Following familiarization to an object undergoing one of two types of motion, visual preferences for the novel motion were assessed after retention intervals of 1 min, 1 day, and 1 month (Experiment 1, N = 120) and 1 min, 1 day, 2 weeks, and 1 month (Experiment 2, N = 74). Results of both studies indicated a significant preference for the novel motion at the 1-min delay, a significant preference for the familiar motion at the 1-month delay, and no preferences at the intermediate retention intervals. In Experiment 3, memory was assessed after a 3-month interval and again, a significant familiarity preference was obtained. These results demonstrate that memory for object motion lasts across retention intervals of 1 and 3 months and that novelty and familiarity preferences interact with retention time. A four-phase function relating visual preferences and retention time was proposed. Phase 1, recent memory, is characterized by a novelty preference; phase 2, intermediate memory, is a period of transition characterized by no visual preference; phase 3, remote memory, is characterized by a familiarity preference; and phase 4, inaccessible memory, is also characterized by no preference. The finding of a transition period at intermediate retention times suggests that null preferences should not necessarily be taken as evidence of forgetting. Rather, more extended retention intervals should be included to interpret null findings obtained in the novelty preference method.