Sixty-three male pigs (Landrace x Large White) weighing 49.5 +/- 0.40 kg were used to (1) examine the variation in DE content of Lupinus angustifolius L. in relation to variety and geographical growing region and (2) establish prediction equations for DE content from physical and chemical composition. The pigs were randomly allocated to a 4 x 2 factorial treatment design with respective factors being 4 varieties (cv. Belara, Coromup, Mandelup, and Tanjil) and 2 growing locations (northern and southern agricultural areas of Western Australia). In addition, a wheat control diet was fed as a reference for calculation of lupin DE content. The lupins were ground through a hammer mill fitted with a 4-mm screen to a mean particle size of 888 mum. Pigs were fed their respective experimental diets at 3 times maintenance energy level [3 x (0.458 x BW(0.75))/diet DE] in the study. The DE content of lupins ranged from 13.3 to 15.7 MJ/kg with a mean value of 14.2 MJ/kg. Variety of lupins affected (P < 0.01) the DE content, and lupins grown in the northern agricultural region had a greater DE content than the same lupins grown in the southern agricultural area (P < 0.01). Although the variation in DE content of lupins was mostly caused by significantly greater DE content of cv. Coromup grown in the northern agricultural region, the results suggest that genetic and environmental conditions during the growth of lupins have a significant impact on the utilization of energy in grower pigs. Simple regression analysis showed that prediction of DE content was possible from the proportion of hulls [R(2) = 0.88, residual SD (RSD) = 1.116, P < 0.001], 1,000-seed weight (R(2) = 0.77, RSD = 1.092, P < 0.01), and soluble arabinoxylan content (R(2) = 0.64, RSD = 1.072, P < 0.05). Multiple regression analysis showed that adding total nonstarch polysaccharide (R(2) = 0.96, RSD = 1.187, P < 0.01) and soluble nonstarch polysaccharide (R(2) = 0.95, RSD = 1.200, P < 0.01) to the equation along with the proportion of hull and 1,000-seed weight significantly improved the accuracy of prediction. Results indicate that the DE content of lupins varies by up to 2.4 MJ/kg and that the DE content can be predicted with a good degree of accuracy using physical and chemical characteristics.