The complaint of knee pain is frequent in children. The difficulty of obtaining an accurate history from the younger child adds to the already-difficult problem of establishing an accurate diagnosis from clinical examination and routine radiographs in this group. This report is a prospective study of 32 children less than 18 years of age who underwent arthroscopy either because their symptoms failed to resolve with conservative treatment or because knee surgery was planned. The findings suggested some important differences between these children over 13 years of age and those age 13 years or younger. In the preadolescent group, 36% gave a history of trauma, 27% were correctly diagnosed before arthroscopy, and the correct diagnosis was delayed for an average of 18 months, as compared to 20%, 61% and nine months, respectively, for the older group. Three major problem areas were identified in the preadolescent group that could be solved by arthroscopy: serious intraarticular damage from long-unrecognized internal derangements; the difficulty of diagnosing the cause of synovitis; and the syndrome of persistent pain with no demonstrable knee pathology. Arthroscopy in the preadolescent patient is a safe and useful tool in those unusual cases with persistent and undiagnosed knee symptoms.