To compare outcomes from pediatric intensive care in hospitals with different levels of resources. Prospective, blinded comparison of outcome and care Tertiary (n = 3) and nontertiary (n = 71) hospitals in Oregon and southwestern Washington. All critically ill children admitted with respiratory failure and head trauma for 6 months. Severity of illness adjusted mortality rates were determined using admission day, physiologic profiles (Pediatric Risk of Mortality score) and care modalities were assessed daily. The crude mortality rate of the tertiary patients was four times higher than for the nontertiary patients (23.4% vs. 6.0%, p < .0001). In the tertiary patients, the numbers of outcomes were accurately predicted by physiologic profiles (observed: 30 deaths and 98 survivors; predicted: 29.3 deaths and 98.7 survivors, z = −.25, p > A). However, for the nontertiary patients, the number of the deaths were significantly different than predicted (observed: 20 deaths and 315 survivors; predicted: 14.4 deaths and 320.6 survivors, z = −2.08, p < .05). The odds ratios of dying in a nontertiary vs. a tertiary facility were about 1.1,2.3, and 8 (p < .05) for mortality risk groups of 30%. Patients in tertiary facilities received more (p < .05) invasive (e.g., arterial catheters) and complex (e.g., mechanical ventilation) care, whereas patients in nontertiary facilities received more (p < .05) labor-intensive care (e.g., hourly vital signs). Care of the most seriously ill children in tertiary pediatric ICUs could improve their chances of survival. (Crit Care Med 1991; 19:150)