- 1. Average daily metabolic rate (ADMR) is the metabolizable energy intake required by a caged animal to maintain constant body energy content. Linear regression of standardized, phylogenetically independent contrasts was used to describe the allometric relationship between ADMR and body mass in rodents. It was predicted that desert species would have a lower ADMR than non-desert species and that granivores would have a lower ADMR than those of herbivores and omnivores. The efficiency of utilization of energy for maintenance was also determined in these species. Here, it was predicted that the efficiency in granivores would be higher than in both herbivores and omnivores and in desert species would be higher than in non-desert species. 2. The regression of the phylogenetic contrasts of ADMR on body mass (mb) for 53 independent data points based on 63 species of rodents was highly significant (P < 0·0001) with an allometric slope of 0·53. Independent comparisons of both basal metabolic rate (BMR) and ADMR showed significantly lower rates in desert than in non-desert species. There was a trend for granivores to have a lower ADMR per g0.53 than either omnivores or herbivores, but this trend is not significant in independent comparisons between the categories. Thus, the predictions in regard to desert and non-desert habitats were confirmed, and there is a non-significant trend in the predicted direction with respect to diet. Herbivores and omnivores showed no significant difference in either BMR or ADMR. 3. In independent comparisons between categories, there was no evidence for trends in the efficiency of utilization of energy for maintenance (BMR:ADMR ratio or km) with either habitat or diet categories. Thus, these results are not in agreement with our predictions. Taking simple means, km equals 0·47±0·21, 0·37±0·12 and 0·56±0·11 for granivores, omnivores and herbivores, respectively, and 0·37±0·12 and 0·55±0·16 for desert and non-desert mammals, respectively. 4. It is recommended that conditions for ADMR measurements be standardized, as is common for basal metabolic rate measurements. This would allow more reliable comparisons among species measured in different laboratories and also provide more accurate estimates of efficiency of utilization of energy.