This paper presents the results of a prospective study of percutaneous transluminal angioplasty (PTA) for the treatment of patients with peripheral arterial occlusive disease and identifies the variables that are predictive of long-term success. The variables believed to be important prognostically were recorded for 984 consecutive PTAs performed between July 1978 and July 1986. Success or failure was defined using a combination of clinical and objective vascular laboratory criteria. The overall long-term success was estimated by the Kaplan-Meier method and differences between curves of success rate versus time for each variable were determined by the Wilcoxon and log-rank statistics. The combination of variables associated with success were determined by the Cox proportional hazards regression model. For all cases, the initial success rate was 88.6 +/- 1.0% and at 5 years was 48.2 +/- 2.3%. The following variables, when considered individually, were associated with success (p less than 0.05): indication for PTA, site of PTA, severity of lesion, runoff, number of sites dilated, diabetes, and the occurrence of a complication. From the Cox model, by using a stepwise multiple regression procedure, the following combination of variables were found to be predictive of success (p less than 0.05): (1) indication (claudication vs. salvage), (2) site (common iliac vs. other), (3) severity of lesion (stenosis vs. occlusion), and (4) runoff (good vs. poor). For all combinations of these four significant variables, curves of the success rate versus time were calculated. In conclusion, this study has identified the combination of four variables that together predict if PTA is likely to be successful in the management of a patient with peripheral arterial occlusive disease.