The roles of ultimate and proximate factors in regulating basal and summit metabolic rates of passerine birds during winter have received little study, and the extent to which winter temperatures affect these variables is unknown. To address this question, we measured basal and summit (maximum cold-induced) metabolic rates in black-capped chickadees (Poecile atricapillus), dark-eyed juncos (Junco hyemalis), and American tree sparrows (Spizella arborea) during winters from 1991/1992 to 1997 in southeastern South Dakota. Both temperature and these metabolic rates varied within and among winters. Least-squares regression revealed significant negative relationships for normalized basal and summit metabolism against mean winter temperature for all species pooled (R2=0.62 to 0.69, P</=0.001). Simple and multiple regressions were performed to analyze the influence of short-term (0-7 d preceding testing), medium-term (14-30 d before testing), and long-term (100-yr means for mean, minimum, and extreme low temperatures) temperature variables on whole-animal and mass-specific metabolic rates. Simple correlation coefficients for whole-animal metabolic rates were highest (r=-0.48 to -0.75, P<0.01) for 14-30-d temperature variables in chickadees and juncos and for 0-5-d temperature variables (r=-0. 54 to -0.68, P<0.01) in tree sparrows. For mass-specific metabolic rates, simple correlation coefficients were again highest (r=-0.45 to -0.70, P<0.01) for 14-30-d temperature variables in chickadees and juncos. Simple correlations for mass-specific metabolic rates were highest for 7-14-d temperature variables for tree sparrows (r=-0.67 to -0.68, P<0.01). Multiple regressions yielded model R2s ranging from 0.45 to 0.94 using a forward selection procedure and from 0.23 to 0.71 using a stepwise selection procedure. The partial R2 contributed from mass variation was small in all cases, ranging from 0.05 to 0.18, indicating that winter temperature was generally a good predictor of metabolic rate in these species. Metabolism was substantially correlated with short- and medium-term temperature variables for all species (cumulative partial R2=0.31 to 0.70 for forward selection and 0.13 to 0.57 for stepwise selection) but, at most, only weakly so with long-term temperature variables (cumulative partial R2=0-0.11 for forward selection and 0-0.06 for stepwise selection). Thus, short- to medium-term temperatures were better predictors of metabolic rates than long-term temperatures. These data suggest a proximate role for winter temperature in regulating metabolism in these birds.