The effect of food on didanosine bioavailability and interpatient pharmacokinetic variability was examined in children infected with human immunodeficiency virus type 1 (HIV-1). Didanosine pharmacokinetics were determined during fasting and fed conditions in HIV-infected children enrolled in the Pediatric AIDS Clinical Trials Group Protocol 144 randomized to receive didanosine at 50 mg/m2 or 150 mg/m2 orally every 12 hr. Pharmacokinetic parameters from patients in the low (n = 39) and high (n = 38) dosing groups were not significantly different, but intersubject variability was substantial. The fraction absorbed was higher while fasting than with food (0.27+/-0.13 versus 0.19+/-0.09, p < 0.0001); the zero order absorption rate was faster (0.48+/-0.31 versus 0.76+/-0.72 hr, p = 0.003); and the plasma half-life was shorter (0.93+/-0.43 versus 1.39+/-0.65 hr, p < 0.0001). The lower fraction absorbed with food was offset by the absorption rate becoming rate limiting for elimination, resulting in similar areas under the concentration-time curves (normalized to 100 mg/m2) when fasted (853.9+/-465.8 microg/liter-hr) versus fed (796.3+/-367.5 microg/liter x hr). Oral clearances during fasting (152.5+/-81.7 liters/hr/m2) and fed states (163.6+/-99.3 liters/hr/m2) were similar, but these values in children are substantially higher than previously reported for adults. The systemic exposure (i.e., AUC) of didanosine was highly variable in children but similar in the presence and absence of food. Administration of didanosine with food in children may be permissible if total systemic exposure rather than maximum plasma concentration is sufficient for antiretroviral activity. The higher oral clearance and substantial pharmacokinetic variability suggest the need to reexamine current didanosine dose recommendations for HIV-infected children.