The disposition of phenylbutazone (4.4 mg/kg), administered intravenously to six Welsh Mountain ponies, was described by a two-compartment open model. Pharmacokinetic parameters were not significantly different after morning dosing in comparison with afternoon dosing. When phenylbutazone (4.4 mg/kg) was administered orally to the same ponies, marked variations in time to peak concentrations were produced with different feeding schedules. When access to hay was permitted before and after dosing, the mean time to peak concentration was 13.2 +/- 1.2 h and double peaks in the plasma concentration-time curve were common. Double peaks were also encountered when phenylbutazone was given to ponies deprived of food prior to, and allowed access to hay after, dosing. In this circumstance, mean times to peak concentration were much shorter (3.8 +/- 1.3 h after morning dosing and 5.3 +/- 1.5 h followed afternoon dosing). Absorption was more regular and double peaks were less apparent when food was withheld both before and after dosing. In order to explain these findings, it is tentatively postulated that, whereas some of the administered dose of phenylbutazone may be absorbed quickly, some may become adsorbed on to the feed and subsequently released by fermentative digestion in the large intestine and/or caecum. The consequences of delayed absorption in fed animals for toxicity and clinical efficacy, and for the use of phenylbutazone in equestrian sports, are considered. Delayed absorption in ponies given access to hay was not accompanied by a significant reduction in total absorption. Bioavailability was estimated to be approximately 69% in fed and 78% in unfed ponies. Estimates of bioavailability gave similar values for morning (72%) and afternoon (71%) dosing.