A retrospective study of 1092 pregnant women showed that the use of a centile chart of maternal weight gain was not effective at detecting women who give birth to small for gestational age (SGA) infants. Multiple regression analysis showed that of the various weight measurements recorded, weekly weight gain and maternal size at booking were the only factors which had a significant association with infant birthweight after taking into account maternal gestation, age, smoking habit and parity. Low maternal booking weight (less than 51 kg) was the most effective maternal weight measurement for antenatal detection of SGA infants (positive predictive value 20.0%). Low average weekly maternal weight gain (less than 0.20 kg) had a positive predictive value of only 12.9% for detecting these pregnancies. Weight loss or failure to gain weight over a 2-week interval occurring in the third trimester was observed in 46% of all women studied. Maternal smoking had a positive predictive value of 16.3% in antenatal detection of SGA infants. Maternal weight need be recorded only at booking, with the exception of patients in whom nutrition is of concern. The subsequent routine weighing of patients may produce unnecessary anxiety and should cease.