To determine the effect of selective termination of an abnormal twin on the rate of preterm delivery. The study group consisted of 69 patients with twin pregnancies who underwent selective termination between 1987–1994. The comparison groups consisted of singleton (n = 42,362) and twin pregnancies (n = 825) delivered at our institution during the same period. The data were analyzed by χ2 and Fisher exact test, as appropriate. P < .05 was considered significant. Adjusted odds ratios (OR) and 95% confidence intervals (CI) were calculated by multiple logistic regression. Terminating the presenting twin was associated with a significantly higher risk of delivery before 37 weeks (adjusted OR 4.1, 95% CI 1.4–12.3) and low birth weight (adjusted OR 3.8, 95% CI 1.3–11.4) compared with terminating the nonpresenting twin. When the termination was performed at or after 20 weeks' gestation, only the risk of preterm premature rupture of the membranes or preterm labor was significantly increased (adjusted OR 3.7, 95% CI 1.2–11.5). Selective termination patients had a lower rate of preterm delivery than twin pregnancies (40 versus 58%, P < .005) but higher than singleton pregnancies (40 versus 10%, P < .001). Selective termination of the presenting twin at 20 weeks or later was associated with a worse perinatal outcome than terminating the nonpresenting twin or performing the procedure before 20 weeks.