This study examines whether patient outcomes are affected by changes in volume over time within hospitals and whether such effects are consistent with cross-sectional results previously reported in the literature. Investigating the existence of volume-outcome relationships longitudinally for specific groups of patients relates directly to the policy issue of whether, and how, specific inpatient services should be regionalized. The analysis uses up to 8 years of observations from a national sample of nearly 500 community hospitals. Outcomes are measured as inhospital mortality adjusted for case severity. Instrumental variables techniques are used to test and control for the possibility of selective referral. The results suggest that higher volume leads to better outcomes for certain groups of patients. Among the groups studied here, increases in volume lowered adjusted mortality rates for acute myocardial infarction, hernia repair, and respiratory distress syndrome in neonates; correlations were observed between volume and outcome for coronary artery bypass grafts, which seemed to be due primarily to referral patterns; and, no significant findings were found for hip replacements. In general, the effects of volume on outcome appear to be larger when estimated from longitudinal, rather than cross-sectional, data.