This paper reports the results of a study of the association between volume of activity and patient outcome in 26 intensive care units in the United Kingdom. Hospital fatality rates showed a statistically significant (p = .016) negative association with volume. However, mean APACHE II scores, as a measure of the severity of case mix, were also negatively associated with volume (p = .021). Thus, one explanation of the lower death rates at higher volumes is that larger units admit less severely ill patients. For patients admitted immediately after surgery, the correlation between severity standardized mortality ratios and volume (while not significant, p < .1) suggests there may be a volume-output effect unexplained by severity.