Recent studies of lactation in wild mammals have emphasised the need for accurate information on milk composition, requiring the analysis of large numbers of samples. A simplified method for determining milk composition in pinnipeds was assessed using samples collected from Antarctic fur seals (Arctocephalus gazella). The stoichiometric relationships between elemental C, H, and N in the ash-free dry mass were used to determine the organic composition and energy content of samples. The results were compared with those obtained by conventional biochemical analyses. Stoichiometric estimates of lipid content were positively correlated with those obtained by methanol–chloroform extraction (r2 = 0.62, P < 0.0001) but consistently lower (P < 0.0005). The mean difference between the results of the two methods was 3.56 ± 0.89% of total milk volume (range −16.98 to +7.47%). Variation between replicates was significantly lower when the stoichiometric method was used (P < 0.05). Gross energy content, calculated from stoichiometric estimates of milk composition, was closely correlated with, though consistently higher than (P < 0.0005), that determined by bomb calorimetry (r2 = 0.96, P < 0.0001). The difference between the two methods, however, was negligible (1.8 ± 0.08%, range −5.0 to +5.6%). The sum of the milk components measured by the stoichiometric method accounted, on average, for 96.5% of the wet mass and varied inversely with lipid content (r2 = 0.29, P < 0.001). This small error is most likely due to retained water in the oven-dried milk. The stoichiometric method for determining organic composition allows the processing of samples at a much faster rate, requiring much smaller volumes, than conventional techniques and should prove useful in ecological studies of lactation.