Maximal transdiaphragmatic (Pdimax) and airway occlusion pressures (PaOmax) have been used to predict weaning from mechanical assisted ventilation in adults, but criteria for weaning are still based on trial and error in infants and young children. Because infants and young children cannot cooperate, crying Pdi and PaOmax against an occlusion have been used, but these may not yield maximal values. We hypothesized that breathing CO2 would achieve better Pdimax and PaOmax values by maximizing respiratory drive and help in establishing weaning criteria. To test this, we measured tidal breathing and occluded Pdi and PaOmax in 27 patients (mean age, 15.0 +/- 31.5 SD months) who required prolonged assisted mechanical ventilation and had failed previous weaning attempts. Measurements were performed while patients were breathing spontaneously 100% O2 and 5% and 7% CO2 in O2. The patients achieved higher Pdimax breathing 5% CO2 (73.2 +/- 24.4 cm H2O) than in O2 (61.6 +/- 24.4 cm H2O; p < 0.0001) or in 7% CO2 (69.1 +/- 23.4 cm H2O; p < 0.0001). They also achieved higher PaOmax in 5% CO2 (81.7 +/- 23.5 cm H2O) than with the other gases (69.9 +/- 25.5 in O2, and 77.5 +/- 24.1 in 7% CO2; p < 0.001); 19 patients (70%) were weaned from assisted ventilation within 3.2 +/- 1.9 wk. In 5% CO2, all patients who were weaned achieved Pdimax > 60 cm H2O and could sustain > 60% Pdimax for more than five successive occluded breaths (100% sensitivity; 100% specificity; p < 0.0005).(ABSTRACT TRUNCATED AT 250 WORDS)