Slow-release theophylline preparations have been prescribed increasingly at the Brompton Hospital, and a serum theophylline assay is requested about once for every ten prescriptions. In a period of eighteen months 1913 such assays were performed, and their influence on management was assessed by retrospective analysis in two groups - 113 outpatients on stable dosage with no recent exacerbations of disease; and 28 inpatients and outpatients with concentrations exceeding 25 mg/l. In those with stable treatment, there was only a loose relation between dose and serum theophylline level. High serum levels were associated with abnormal liver function, diuretic use, and duplicate prescribing. When serum theophylline level were reported high, dosage was always reduced, whereas a low serum theophylline prompted dosage increase in only one third of patients. Of the patients with very high serum theophylline concentrations (over 25 mg/l), 18 had been receiving oral theophylline and 10 aminophylline by infusion. 3 had seizures, of whom 2 died, and 1 patient was resuscitated after cardiac arrest. Unsuspected theophylline toxicity may be responsible for considerable hidden morbidity.