A large multicentre study on the short and long term clinical and developmental outcome of infants randomised to different diets is being undertaken. This report represents an interim analysis of the early postnatal growth performance of an unselected population of 194 preterm infants (gestation, mean (SD) 31 . 0 (2 . 9) weeks; birthweight, mean (SD) 1364 (294) g), both ill and well, examined in two (of four) parallel trials. One trial compared banked breast milk with a new preterm formula (primary trial); the other compared these diets as supplements to maternal milk (supplement trial). A major dietary effect on the number of days taken to regain birthweight and subsequent gains in weight, length, and head circumference was observed in the primary trial. Infants fed banked breast milk and weighing less than 1200 g at birth took a calculated additional three weeks to reach 2000 g compared with those fed on the preterm formula. A significant influence of diet on body proportions was seen in the relation between body weight, head circumference, and length. Similar though smaller differences in growth patterns were seen in the supplement trial. By the time they reach 2000 g, infants of birthweights 1200 to 1849 g fed on banked breast milk and infants below 1200 g fed on either banked breast milk or maternal milk supplemented (as necessary) with banked breast milk, fulfilled stringent criteria for failure to thrive (weight less than 2 SD below the mean for age). Only infants fed the preterm formula as their sole diet had maintained their birth centile by discharge from hospital. The misleading nature of comparisons between extrauterine and intrauterine steady state weight gains is emphasised.