To evaluate the impact of patient, tumor, and treatment-related factors on outcome in unselected patients with recurrent osteosarcoma. Five hundred seventy-six consecutive patients who had achieved a first complete surgical remission (CR) during combined-modality therapy on neoadjuvant Cooperative Osteosarcoma Study Group (COSS) protocols and then developed recurrent osteosarcoma were analyzed (median time from biopsy to relapse, 1.6 years; range, 0.1 to 14.3 years). There were 501 patients with metastases, 44 with local recurrences, and 31 with both. Metastases involved lungs (469 patients), bones (90 patients), and/or other sites (54 patients). After a median follow-up of 1.2 years for all patients and 4.2 years for survivors, actuarial overall survival (OS) rates at 2, 5, and 10 years were 0.38, 0.23, and 0.18, respectively. Five-year OS was 0.39 for 339 patients with and 0.00 for 229 patients without a second surgical CR (P < .0001). A long time to relapse, a solitary lesion, and, in the case of pulmonary metastases, unilateral disease and the absence of pleural disruption, were of positive prognostic value in uni- and multivariate analyses, as were a second surgical CR and the use of second-line chemotherapy. Radiotherapy was associated with moderately prolonged survival in patients without a second CR. The very limited prognostic differences associated with the use of second-line chemotherapy appeared to be more pronounced with polychemotherapy. Time to relapse and tumor burden correlate with postrelapse outcome in osteosarcoma. Complete surgery is an essential component of curative second-line therapy. Chemotherapy, particularly chemotherapy with more than one agent, may contribute to limited improvements in outcome.