Four- and 5-year-olds' understanding of basic turtle graphics commands was examined before and after a hands-on, interactive problem-solving experience. Children (n=32) saw display screen events consisting of an initial turtle state, a command transformation, and the resulting turtle state. They were asked to give the command executed in each event. Most children systematically misunderstood some or all of the commands, and their misconceptions were in line with Piaget's characterization of young children's thinking. Contrary to anecdotal evidence, misconceptions were affected little by experience, feedback, and eventual success in solving turtle graphics problems. (Author/RH)