We have constructed an automated, eight-cage indirect calorimeter (AIC) for the measurement of energy expenditure in rats. We compared the measurements of resting energy expenditure (REE) in rats during a 30-h fast obtained with the AIC with those obtained with a manual indirect calorimetry (MIC) system. There was both a high degree of correlation between the two techniques during the initial 18 h of the fast (r = 0.90, P < 0.05) and strong intertechnique agreement. REE (AIC) decreased during the final 12 h of the 30-h fast (79.6 +/- 2.7-72.0 +/- 4.4 kcal.kg-0.75.d-1 [mean +/- SD, P < 0.01]). REE (MIC) did not show a significant decrease during this part of the fast (79.7 +/- 2.6 - 75.2 +/- 4.7 kcal.kg-0.75.d-1 [P = NS]). During the final 12 h of the fast agreement between the two systems gradually dissipated and correlation was poor (r = 0.375, P < 0.05). The frequency of animal handling necessitated by MIC may have resulted in a stress-induced increase in metabolic work that would mask the animals' adaptive response to starvation. This investigation demonstrates the advantages of the AIC and calls into question the accuracy of manual methods under long-term starvation conditions.