Previous studies have shown that people's ability to detect, from memory, alterations in highly familiar faces is excellent. Indeed, just noticeable differences for the detection of small alterations in a recognition-memory task were not significantly different from the corresponding measures in a perceptual-discrimination task (Brédart and Devue, 2006 Perception 35 101-106; Ge et al, 2003 Perception 32 601-614). The object of the present study was to evaluate whether people's perceptual memory for body shapes of very familiar persons reaches the high level of precision that was reported for face memory. For one group of participants, the task was to detect body shape alterations (an increase or a decrease of 2% to 10% of the waist-to-hip ratio) on photographs depicting either themselves or a friend. For another group of participants who did not know the target persons, the task was to discriminate whether two photographs presented side by side were the same or not. Results showed that the detection of alterations was significantly better in the perceptual-discrimination task than in the recognition-memory tasks (for the participant's own body as well as for the friend's body). In conclusion, the high fidelity of perceptual memory for very familiar faces does not extend to familiar body shapes.