Purpose-We have previously reported wide variations among urologists in use of androgen deprivation for prostate cancer. Using Surveillance, Epidemiology and End-Results (SEER)-Medicare linked data we examined how individual urologist characteristics influenced use of androgen deprivation therapy.Methods-Participants included 82,375 men with prostate cancer diagnosed from January 1, 1992, through December 31, 2002, and 2,080 urologists providing care to them. Multi-level analyses were used to estimate likelihood of androgen deprivation use within six months of diagnosis in the overall cohort, a subgroup where use would be of uncertain benefit (primary therapy for localized prostate cancer), and a subgroup where use would be evidence-based (adjuvant therapy with radiation for locally advanced disease).Results-In the overall cohort of patients, a multi-level model adjusted for patient, tumor and urologist characteristics (board certification, academic affiliation, patient panel size, years since medical school graduation) showed that the likelihood of androgen deprivation use was significantly higher for patients who saw urologists without an academic affiliation. This pattern was also noted when the analysis was limited to settings where androgen deprivation would have been of uncertain benefit. Odds ratios for use in that context were 1.66 (95%CI: 1.27-2.16) for no academic affiliation and 1.45 (95%CI:1.13-1.85) for minor versus major academic affiliation.Conclusion-Use of androgen deprivation for prostate cancer varies by the characteristics of the urologist. Patients of non-academically affiliated urologists were significantly more likely to receive primary androgen deprivation therapy for localized prostate cancer, a setting where the benefits are uncertain.