Numerous physiologic factors affect the disposition of theophylline. One nonpatient factor that can influence steady-state theophylline concentration is the administered dosage. The accuracy and variability of hospital pharmacy-prepared i.v. admixtures of theophylline has not been quantitated. A study was designed to evaluate variation in theophylline concentrations from two sources of theophylline admixtures--one by a hospital pharmacy i.v. room and one by a pharmaceutical manufacturer making a premade product. For the theoretical 1.6 mg/ml admixture, the mean theophylline concentration of the pharmacy-prepared solution was lower than that of premixed, whereas the absolute percent error of the premixed product was less than that of the pharmacy bags. For the theoretical 3.2 mg/ml admixture, the theophylline concentration of the premixed product was lower than that of both pharmacy products, whereas the absolute percent error of the pharmacy bags was less than that of the pharmacy bottles and premixed bags. Our data imply that variability in theophylline concentration can occur depending on the method of preparation, drug concentration, and formulation. Pharmacokinetic monitoring of theophylline should include an assessment of methods of i.v. drug preparation. Pharmacy departments should have a policy that assures consistency in the method of preparation of i.v. drugs.