To evaluate the problem-solving abilities of a cohort of inner-city cocaine-exposed children and controls, children were invited to play with the Goodman Lock Box, a large red and blue box with 10 compartments, each having a different lock and toy. Examiners, blinded to the children's group status, coded the children's activities during the 6.5-minute play period. Nineteen behaviors were collapsed into three outcomes: Aimless Actions, Competence, and Mental Organization. Groups' scores were compared with scores of the Goodman standardization sample of mixed socioeconomic status preschoolers. Seventy-three cocaine-exposed children and 82 controls were evaluated at age 3.5 years; of these, 58 cocaine-exposed and 63 controls were reevaluated at age 4.5 years. The groups' scores did not differ on any outcome at either time point (p > or = .22). However, both groups' Mental Organization scores were consistently lower than the Goodman group at each age (p < .01). This high-risk cohort may experience problems functioning in more complex environments such as the classroom.