This study delineates the roles of "selective referrals" and "practice makes perfect" in the hospital procedure volume and in-hospital mortality association for radical cystectomy and radical prostatectomy. This is a retrospective analysis of the Nationwide Inpatient Sample (years 2000-2004). All hospitalizations with primary procedure codes for radical cystectomy and radical prostatectomy were selected. The association between hospital procedure volume and in-hospital mortality was examined using generalized estimating equations and by instrumental variables approaches. There was an inverse association between hospital procedure volume and in-hospital mortality for radical cystectomy (odds ratio = 0.57; 95% confidence interval = 0.38-0.87; P < .05). Results from the 2-stages least squares regression approach suggested that receiving treatment in high-volume hospitals decreased the probability of in-hospital mortality by 0.02 points, compared with 0.01 points using the ordinary least squares regression approach. Outcomes following radical cystectomy appear to be driven by "practice makes perfect."