An analysis of the faecal pellets of two species of arid-zone mouse, the house mouse (Mus domesticus) and Lakeland Downs short-tailed mouse (Leggadina lakedownensis), inhabiting Thevenard Island in Western Australia was conducted to ascertain their dietary requirements, and to use this information to predict their feeding rates in the field. Both species consumed seed, monocotyledon and dicotyledon plant material and invertebrate material, although the relative frequency-of-occurrence of these items varied throughout the year. Invertebrate material formed the highest proportion of dietary intake for both rodent species at all times, suggesting that this dietary strategy is advantageous for rodent species that inhabit environments where plant seeding is seasonal and rainfall dependent. The dry-matter intake (DMI) of free-ranging M. domesticus and L. lakedownensis was predicted and compared using information from two sources: their isotopic water fluxes and the water content of their diet, and their isotopic sodium fluxes and the sodium content of their diet with and without corrections for non-dietary (exogenous) sources of sodium. The DMI derived from the water turnover was high for both species, suggesting that the mice were drinking and that assumptions inherent in this calculation were violated. Feeding rates were also high if no correction was made for exogenous, non-dietary sodium. When corrections were made, however, M. domesticus was predicted to ingest 4.62 +/- 0.20 g dry matter day(-1) compared with 3.86 +/- 0.23 g dry matter day(-1) for L. lakedownensis. When DMI was scaled on the basis of allometric predictions for desert eutherians, only estimates of DMI for M. domesticus fell outside the predicted 95% confidence intervals. The results presented suggest that M. domesticus were obtaining some sodium from sources additional to their diet. Taken together, this methodology provides a useful application for measuring the feeding rate of free-ranging species given known dietary requirements in the field.