Sixty 4- and 6-year-old children observed a female model consistently choose the items of one category with either positive (PVR), negative (NVR), or no (NC) reinforcement for her choice responses. Following this, they were given a performance and an acquisition test. The results showed that while the older children imitated the model's response on the three vicarious reinforcement conditions differently, the younger children imitated the model to the same degree in PVR and NC conditions, which were both superior to NVR condition. The older children made significantly more matching responses to the model than the younger children in the PVR condition, but the reverse was true in the NVR condition. The results were discussed in terms of the developmental difference in the mode of processing information from vicarious reinforcement.