Meta-analysis was used to evaluate the response of dairy cows to different stressors. Holstein cows (n = 24; 2.8 ± 0.7 parity; 153 ± 25 DIM) were housed in free-stalls for 14d (P1), then moved to environmentally controlled tie-stall rooms at thermoneutral temperature for 7d (P2; constant 20°C), then exposed to 14d of programmed 12h cyclical heat stress (HS; mean THI range 72.5 ± 0.1 to 81.3 ± 0.5; P3), and finally, moved back to free-stalls for 7d (P4). Response of animals to housing (P1 vs. P2), HS (P2 vs. P3), and recovery from HS (P3 vs. P4; effect confounded with housing) were evaluated. For analyses, each variable within each period was averaged over the 3 median days. Data were analyzed using Proc MIXED of SAS; the original treatments were random effects and period was a repeated effect. For each comparison, data from the previous period were used to covariate-adjust for carryover effects. Moving cows from free- to tie-stalls did not affect BW, but reduced (P < 0.01) milk yield and DMI by 5 and 10%, respectively. Feed efficiency (FE) and ECM FE improved 6% (P < 0.03) but milk components declined (P < 0.05) from free- to tie-stalls, except for fat and SCS, which increased (P < 0.05). Plasma NEFA decreased (P < 0.01) and BHBA increased (P < 0.03) in tie-stalls. Immune measurements were minimally affected. Respiration rate (RR) increased (P < 0.06) 10% in cows housed in tie-stalls, but rectal temperature (RT) was only 0.2°C higher (P < 0.01). Heat stress negatively affected (P < 0.01) BW, milk yield and DMI (10, 28 and 28% decline, respectively) but did not affect FE, ECM FE or 3.5% FCM FE. Immune response was negatively affected (P < 0.05) during HS. Rectal temperature and RR increased (P < 0.01) 1.5°C and 65%, respectively, in P3 compared with P2. During recovery from HS (P4), milk yield and DMI were higher (P < 0.03), FE and ECM FE were lower (P < 0.01), and BW did not change from P3. Although recovery from HS markedly reduced RT and RR (1.7°C and 42%, respectively; P < 0.01), milk composition and most immune responses did not change from P3 to P4. We conclude that different stressors elicit different responses in performance variables and immune function.