OBJECTIVE To determine the analytic advantages obtained from separating unmarried mothers with partners from unmarried mothers without partners when assessing risk of adverse birth outcomes. METHODS Data were obtained from Listening to Mothers II, a national survey of women's childbearing experiences. Marital status was asked with three choices: married (71%), unmarried with partner (24%), and unmarried without partner (5%). Demographic differences between the three marital status groups were compared using chi (2) tests. Multiple logistic regressions, controlling for age, education, race/ethnicity, and parity, tested for associations between birth outcomes (birth weight and gestational age) and marital status. RESULTS Unmarried mothers with partners and without partners were similar in age, education, and parity. Unmarried mothers without partners delivered by cesarean more often (39%) and were more likely to have a doctor as birth attendant (99%) than unmarried mothers with partners. The multiple logistic regressions indicated that, compared to married mothers, unmarried mothers with partners had the same risk of premature infants, while unmarried mothers without partners had greater risk. Unmarried first-time mothers with partners had over twice the risk of premature infants (OR = 2.71; 1.07-6.85) and unmarried mothers without partners had over 5 times the risk (OR = 5.64; 1.68-18.92) when compared to married first-time mothers. CONCLUSIONS Mothers without partners were at higher risk than unmarried mothers with partners indicating a gradient of risk. Future data collection on marital status should consider distinguishing between unmarried mothers with and without partners.