Seventy-one patients with 83 septic joints due to idiopathic septic arthritis were investigated retrospectively. Forty-three joints were closed primarily while 40 underwent delayed primary closure. Two joints in each group failed treatment and had to be reopened after definitive closure. Joints in patients 7 years of age or older had a substantially higher failure rate (12%), with either primary or delayed closure, than joints in patients younger than 7 years (1.7%). Two of 13 patients with a polymorphonuclear leukocyte count in the synovial fluid greater than 100 000 required repeat debridement after final closure. Three of the four patients who required repeat debridement showed evidence of osteomyelitis. The average length of stay was longer in the delayed primary closure group (7.0 days) than in the primary closure group (5.6 days). This study suggests that primary closure is a viable alternative to delayed primary closure with shorter hospital stays and similar outcomes in children with idiopathic septic arthritis. Care should be taken in children aged 7 and older or when concurrent osteomyelitis exists.