A shock canine pneumonia model that permitted relief of discomfort with the use of objective criteria was developed and validated. After intrabronchial Staphylococcus aureus challenge, mechanical ventilation, antibiotics, fluids, vasopressors, sedatives, and analgesics were titrated based on algorithms for 96 h. Increasing S. aureus (1 to 8 x 10(9) colony-forming units/kg) produced decreasing survival rates (P = 0.04). From 4 to 96 h, changes in arterial-alveolar oxygen gradients, mean pulmonary artery pressure, IL-1, serum sodium levels, mechanical ventilation, and vasopressor support were ordered based on survival time [acute nonsurvivors (< or =24 h until death, n = 8) > or = subacute nonsurvivors (>24 to 96 h until death, n = 8) > or = survivors (> or =96 h until death, n = 22) (all P < 0.05)]. In the first 12 h, increases in lactate and renal abnormalities were greatest in acute nonsurvivors (all P < 0.05). Compared with survivors, subacute nonsurvivors had greater rises in cytokines and liver enzymes and greater falls in platelets, white cell counts, pH, and urine output from 24 to 96 h (all P < 0.05). Importantly, these changes were not attributable to dosages of sedation, which decreased in nonsurvivors [survivors vs. nonsurvivors: 5.0 +/- 1.0 vs. 3.8 +/- 0.7 ml x h(-1) x (fentanyl/midazolam/ medetomidine)(-1); P = 0.02]. In this model, the pain control regimen did not mask changes in metabolic function and lung injury or the need for more hemodynamic and pulmonary support related to increasing severity of sepsis. The integration into this model of both specific and supportive titrated therapies routinely used in septic patients may provide a more realistic setting to evaluate therapies for sepsis.