Several recent studies have explored gender differences in medical care that are not attributable to clinical characteristics. At an 880-bed teaching hospital between July 1987 and June 1990, we studied the importance of gender on two measures of hospital care: length of stay and ancillary service use. The latter was measured on a relative value unit (RVU) scale, based on an estimation of direct cost dollars. Neither mean age nor in-hospital mortality differed between the 9,102 women and 10,285 men. After case-mix adjustment, women stayed in the hospital 0.22 days longer than men (p = 0.01) but consumed 67 fewer RVUs (p = 0.01). This RVU difference dissolved when intensive care unit (ICU) stays were eliminated; men were 1.13 times more likely (95% confidence interval 1.07 to 1.19) to be placed in the ICU. Being married shortened length of stay and women were less likely to be married (51% vs. 68%; p < 0.001), but even within marital status subgroups women remained in the hospital longer than men. Whether this longer length of stay and less technologically intensive care for women reflects a difference in illness severity or physician gender bias requires further study.