OBJECTIVES: To identify independent risk factors for renal impairment in HRPC patients receiving ZA, we compared logistic regression and Cox proportional hazards models. METHODS: A comprehensive medical record review was performed, using electronic databases and paper records, in a large tertiary oncology center. Inclusion criteria: = 18 years, actively treated, HRPC with BM, at least one ZA infusion (from 12/1999 to 4/2005), and at least one creatinine reading before and after first ZA infusion. Renal impairment was defined as an increase of = 0.5 mg/dL and = 1.0 mg/dL over baseline serum creatinine if baseline was < 1.4 mg/dL and = 1.4 mg/dL, respectively; or any doubling of baseline creatinine. Risk factor analysis was by logistic regression with time adjustment, and by Cox model for time-related binary outcome data. RESULTS: Among the 122 eligible patients (mean age = 70.1), mean ZA treatment lasted 367.2 days (mean 10.7 infusions per patient). About 59% of patients discontinued ZA; 21% due to renal complications. Twenty-nine patients (23.8%, 95% CI: 16.2-31.3%) had renal impairment during treatment; this is higher than previously reported in clinical trials. Renal risk increased with extended ZA therapy (<6 months: 22.5%; =12 months: 23.5%; = 24 months: 31.3%) and previous pamidronate treatment (45.8%). Risk factor Stepwise logistic regression Stepwise Cox regression Increasing age at ZA initiation OR=1.06 per additional year, p=0.0467 HR=1.06 per additional year, p=0.0202 Prior pamidronate use OR=3.36, p=0.0311 HR=2.16, p=0.0516 History of hypercalcemia OR=7.36, p=0.0504 HR=4.26, p=0.0225 Treatment with anti-hypertensives OR=5.22, p=0.0112 HR=3.34, p=0.0258 --> were identified. CONCLUSION: Although regression coefficients are different, these risk factors proved significant in both multivariate models, further supporting that these factors are likely important in predicting the renal impairment associated with ZA use in