Abstract
Skinfold thickness was measured at five sites in 750 infants. The study population was unselected except that twin pregnancies and the infants of diabetic mothers were excluded, and very preterm infants were under represented. A pilot study had indicated that skinfold measurement was most reproducible at the thigh site. Thigh skinfold correlated better with the sum of other skinfolds than did skinfold measurement at any other site and closely resembled the summed skinfold in correlations with a number of maternal and fetal variables. Median skinfold increased with birthweight and was greater in girls than in boys. 'Corrected skinfold', a mathematical approach to comparing skinfolds in infants of differing sex and birthweight, is suggested as an alternative to absolute skinfold measurement.