In 2 experiments, an error-detection approach was used to determine whether 3-year-olds' perseverative errors on the postswitch phase of the Dimensional Change Card Sort (DCCS) are due to lack of response control or representational inflexibility. In Experiment 1, 3-, 4-, and 5-year-olds watched a puppet sort perseveratively on the postswitch phase and evaluated its responses. Most 4- and 5-year-olds detected the puppet's perseverative errors, whereas most 3-year-olds failed to do so despite detecting errors on a simpler card sort. Experiment 2 revealed that 3-year-olds who failed to correctly evaluate the puppet's behavior tended to fail their own DCCS. Results imply that perseveration on the DCCS cannot be attributed to difficulty inhibiting prepotent motor responses. Instead, changes in rule use between 3 and 5 years of age are interpreted in terms of the development of representational flexibility.