Between July 1, 1983 and December 31, 1990, risk factors were determined for all horses with joint disease presented to a referral center, of being discharged, of ever becoming sound, or of being alive at 3 mo follow-up. Logistic multiple-regression models were done separately for foals (< or = 4 mo), yearlings (> 4-24 mo) and racing or nonracing adult horses (> 24 mo). The breakdown in this study was 53 foals, 87 yearlings, 141 nonracing adults, and 226 racing adults. Thirty-one foals (58%), 68 yearlings (78%), 119 non-racing adults (84%), and 213 racing adults (94%) were discharged. Foals with a less severe lameness, duration of illness of > 1 d, and infectious arthritis had increased odds of discharge. At follow-up, 12 of 18 (67%) were alive, 10 (56%) of which were sound. Yearlings with osteochondrosis had higher odds of discharge; at follow-up, 38 of 49 (78%) were alive, 32 (65%) of which were sound. For non-racing adults, horses with less severe lameness, without a miscellaneous diagnosis, or intended for pleasure use had increased odds of discharge. At follow-up, 55 of 78 (70%) were alive and 33 of 58 (57%) with soundness data became sound. Risk factors for higher odds of being alive at follow-up were carpal lameness, arthroscopic surgery, a prognosis other than poor, became sound, above-median hospitalization costs, and duration of follow-up. The 161 racing adults (76% of discharges), with follow-up, were more likely to have had osteoarthritis, higher hospital costs, hospitalization > 1 d, and arthroscopy. Sixty-four (60%) of these became sound; the odds increased if the horse was not severely lame at admission or was hospitalized for > 1 d. Risk factors and prognosis differed by age-use group among horses seen at our hospital.