Predictive relations between maternal variables and outcome status for 199 pairs of newborn twins were evaluated. The maternal independent variables included pregravid weight, weight at end of pregnancy, weight gain, age, parity, height, socioeconomic status, years of education, and body mass index. The twin dependent variables included birthweight, birth length, 5-min Apgar score, number of days in an isolette, number of days in the hospital, weight at discharge, and chronological age at discharge. For stepwise multiple linear regression analyses, the newborn twins were grouped by term (fullterm/preterm) and sex (female/male). A combination of a larger number of independent variables acted as statistically significant predictors of twin outcome for the preterm twins than for the fullterm twins, and for the male twins than for the female twins. This was noted particularly for the anthropometric variables and for parity. Maternal weight gain, typically considered to be the critical variable related to infant birthweight, consistently had a significant outcome role for the preterm twins and male twins, but not for the fullterm twins or female twins. Demographic variables had a low rate of prediction for newborn outcome. This study has demonstrated that evaluating twins within groups separated by potential risk variables allows for a more accurate description of the influence of maternal variables on twin outcome than evaluating twins across groups as a single population.