To clarify the relative importance of clinicopathological factors affecting survival in patients with renal cell carcinoma, univariate and multivariate analyses by Cox's proportional hazards model were performed for 121 patients undergoing nephrectomy between 1980 and 1991. The 5-year survival rate was 67% for all 121 patients. Univariate analysis revealed that distant metastasis, local invasion, venous involvement, infiltration pattern, grade, lymph node metastasis, sex, and tumor size were significantly associated with patient survival. Multivariate analysis using a method of stepwise selection revealed that presence or absence of distant metastasis is the most significant determinant (p < 0.0001) for survival, followed by venous involvement (p < 0.001), treatment period (p < 0.02) and local invasion (p < 0.02), in this order. A four-factor model of the above determinants yielded adjusted hazard ratios of 5.3 for distant metastasis (positive vs. negative), 3.7 for venous involvement (pV1a-pV2 vs. pV0), 3.9 for treatment period (1980-1984 vs. 1985-1991), and 3.1 for local invasion (pT3-pT4 vs. pT1-pT2). The present study revealed recent improvements in the patient survival and justified the clinical application of Robson's staging system implying local invasion, venous thrombus formation and distant metastasis as prognostic determinants.