A computer-assisted analysis of perinatal variables wasundertaken in order to characterize the growth-discordanttwin. We studied 124 twin pairs grouped at four degrees ofdiscordancy: over 5, 10, 15, and 20% birth weight difference, using the larger twin as 100%. Twins with birth weightdiscrepancy below these figures were considered concor-dant. Highly discordant second twins showed significantlylower birth weight, gestational age, and Apgar scores, andprolonged neonatal hospitalization, as compared with discordant first twins. Complex presentations and a highercesarean section rate characterize discordant first and con-cordant twins. It is suggested that a 15% birth weightdiscordancy be used as the safety limit for birth weightdisparity. We also recommend special attention when dis-cordancy seems to increase, especially if the discordant isthe second twin.