Heat stress is detrimental to dairy production and affects numerous variables including feed intake and milk production. It is unclear, however, whether decreased milk yield is primarily due to the associated reduction in feed intake or the cumulative effects of heat stress on feed intake, metabolism, and physiology of dairy cattle. To distinguish between direct (not mediated by feed intake) and indirect (mediated by feed intake) effects of heat stress on physiological and metabolic indices, Holstein cows (n = 6) housed in thermal neutral conditions were pair-fed (PF) to match the nutrient intake of heat-stressed cows (HS; n = 6). All cows were subjected to 2 experimental periods: 1) thermal neutral and ad libitum intake for 9 d (P1) and 2) HS or PF for 9 d (P2). Heat-stress conditions were cyclical with daily temperatures ranging from 29.7 to 39.2 degrees C. During P1 and P2 all cows received i.v. challenges of epinephrine (d 6 of each period), and growth hormone releasing factor (GRF; d 7 of each period), and had circulating somatotropin (ST) profiles characterized (every 15 min for 6 h on d 8 of each period). During P2, HS cows were hyperthermic for the entire day and peak differences in rectal temperatures and respiration rates occurred in the afternoon (38.7 to 40.2 degrees C and 46 to 82 breaths/min, respectively). Heat stress decreased dry matter intake by greater than 35% and, by design, PF cows had similar reduced intakes. Heat stress and PF decreased milk yield, although the pattern and magnitude (40 and 21%, respectively) differed between treatments. The reduction in dry matter intake caused by HS accounted for only approximately 35% of the decrease in milk production. Both HS and PF cows entered into negative energy balance, but only PF cows had increased (approximately 120%) basal nonesterified fatty acid (NEFA) concentrations. Both PF and HS cows had decreased (7%) plasma glucose levels. The NEFA response to epinephrine did not differ between treatments but was increased (greater than 50%) in all cows during P2. During P2, HS (but not PF) cows had a modest reduction (16%) in plasma insulin-like growth factor-I. Neither treatment nor period had an effect on the ST response to GRF and there was little or no treatment effect on mean ST levels or pulsatility characteristics, but both HS and PF cows had reduced mean ST concentrations during P2. In summary, reduced nutrient intake accounted for just 35% of the HS-induced decrease in milk yield, and modest changes in the somatotropic axis may have contributed to a portion of the remainder. Differences in basal NEFA between PF and HS cows suggest a shift in postabsorptive metabolism and nutrient partitioning that may explain the additional reduction in milk yield in cows experiencing a thermal load.