We sought to examine racial/ethnic differences in deliveries by caesarean section (CS) over time, particularly among women at low risk for this procedure. To do so, we conducted a retrospective cohort study at the University of California, San Francisco, a tertiary care academic centre. Births occurring between 1980 and 2001 were included in the analyses. Women with multiple gestations, fetuses in other than the cephalic presentation or with other known contraindications to vaginal birth were excluded. A total of 28 493 African American, Asian, Latina and White women were studied. Risk-adjusted models were created to explore differences in CS risk by race/ethnicity. We also performed analyses of subgroups of women at relatively low risk of CS, and explored changes in observed disparities over time. The overall CS rate was 15.8%. The absolute rate was highest among Latinas (16.7%) and lowest among Asians (14.7%). After adjustment for known risk factors, African American women had a 1.48 times greater odds of having a CS than did White women [95% confidence interval (CI) 1.31, 1.68], and Latina women had a 1.19 times greater odds [95% CI 1.05, 1.34]. Stepwise adjustment for confounders showed that this variation is not entirely explained by known risk factors. These differences exist even for women at low risk of CS, and have persisted over time. We conclude that racial and ethnic disparities in CS delivery exist, even among women presumed to be at lower risk of CS; rates have not improved with time. Disparities in risk-adjusted CS should be considered as a quality metric for obstetric care, whether at the national, state, hospital or provider level.