Detailed studies of energy/nutrient acquisition and food selection have been completed for both African apes and orang-utans in mast-fruiting dipterocarp forests, but, to date, no studies of this type have been completed on orang-utans in non-masting forests, which experience more muted fluctuations in fruit availability. Such studies may be instructive in understanding whether the observations on orang-utans in masting habitats are specific to orang-utans, or specific only to orang-utans in masting habitats. To investigate this, orang-utan diet composition, energy intake, food selection and behaviour were studied in relation to orang-utan fruit/flower availability, and chemical and non-chemical food properties, in the Sabangau peat-swamp forest, Central Kalimantan from July 2005-June 2007. All data were collected using standard methods, and comparisons made with published data on orang-utans in mast-fruiting habitats and on African apes. Orang-utan fruit availability in Sabangau was lower and less variable than in the masting forests of Gunung Palung National Park, West Kalimantan. Daily energy intake was below requirements in most months, and similar to periods of very low food availability between mast-fruiting events in Gunung Palung. Mechanisms through which orang-utans may have survived this prolonged energetic shortfall are suggested. In addition, unlike in Gunung Palung, where energy intake and fruit availability are tightly linked in both sexes, the only age-sex class in which energy intake in Sabangau was related to fruit availability was flanged males. Although selection of preferred vs. fall-back food types in Sabangau appears to be governed at least partially by the expected rate of energy returns, fruit and flower selectivity rank appears better explained by food quality: the best multiiii variate model of adult female fruit selectivity rank contained only protein/fibre ratio, and bivariate correlations between fruit-pulp selectivity rank and protein/fibre ratio were also significant for flanged males. Furthermore, adult female and flanged male fruit selectivity rank was negatively affected by fibre, weight or energy intake in at least one bivariate analysis. This is suggested to represent selection against ingestion of large amounts of relatively indigestible fibre, and implies that maximising dietary quality, rather than energy intake, as has been suggested in masting forests in Borneo, may be the optimal feeding strategy for orang-utans in habitats where fruit quality is relatively poor and/or fruit availability is relatively consistent. Finally, compared to orang-utans in Bornean masting forests, Sabangau orang-utans exhibit a relative lack of modifications in nonfeeding behaviours in response to reductions in preferred food availability. These results suggest that, compared to masting forests in Borneo, lower mean quality and quantity of fruit, and lower variability in these parameters, in the Sabangau peat swamps leads to important differences in feeding behaviour between these habitat types. When compared to data on orang-utans in other sites and African great apes, these observations are compatible with a graded-response hypothesis, in which the less predictable the availability of high-energy fruit in a site: (1) the stronger the relationship between fruit availability, fruit consumption and energy intake, (2) the more food is selected based on energy content, and (3) the stronger the influence of fluctuations in fruit availability on behaviour and, ultimately, probably also female reproduction. The implications of these findings are discussed.