To utilize United States birth certificate data (years 1989-1991) to examine the effects of race/ethnicity on twinning rates. We used a database of birth certificate data for the United States (years 1989-1991) available on CD-ROM from the United States National Center for Health Statistics. This linked birth/infant death data set included a total of 12,036,234 birth records, of which 279,073 were twins. Excluded from the analysis were 15,086 twin birth records (5.4%) based upon previously described exclusion criteria. Statistical analysis included a series of univariate analyses to determine the rates of twinning between different racial/ethnic groups correlated with maternal age, education, marital status and place of birth. Comparisons were analyzed using the chi 2 test, with significance at P < .05. The overall twinning rate was 2.26%, with the lowest rate observed in the group denoted non-Hispanic other (1.67%) and highest among non-Hispanic blacks (2.69%). Twinning rates within various groups representing Hispanics ranged from 1.76% to 2.08%. The twinning rate was significantly higher among U.S.-born as compared to non-U.S.-born women (2.34% vs. 1.88%, P < .005) except for Puerto Ricans, Cubans and Central/South Americans, for whom the reverse was true. Women aged 35-39 had the highest twinning rate (3.05%), whereas women aged 15-19 had the lowest; that trend persisted in all racial/ethnic groups. The study showed a nonsignificant relationship between marital status or education and twinning rate. Race/ethnicity studies can be useful in designing programs that may maximize health outcomes of twins in a racially diverse population.