The use of the iron chelator deferiprone (L1, CP20, 1,2-dimethyl-3-hydroxypyrid-4-one) for the treatment of diseases of iron overload and other disorders is problematic and requires further evaluation. In this study the efficacy, toxicity and mechanism of action of orally administered L1 were investigated in the guinea pig using the carbonyl iron model of iron overload. In an acute trial, depletion of liver non-heme iron in drug-treated guinea pigs (normal iron status) was maximal (approximately 50% of control) after a single oral dose of L1 of 200 mg kg-1, suggesting a limited chelatable pool in normal tissue. There was no apparent toxicity up to 600 mg kg-1. In each of two sub-acute trials, normal and iron-loaded animals were fed L1 (300 mg kg-1 day-1) or placebo for six days. Final mortalities were 12/20 (L1) and 0/20 (placebo). Symptoms included weakness, weight loss and eye discharge. Iron-loaded as well as normal guinea pigs were affected, indicating that at this drug level iron loading was not protective. In a chronic trial guinea pigs received L1 (50 mg kg-1 day-1) or placebo for six days per week over eight months. Liver non-heme iron was reduced in animals iron-loaded prior to the trial. The increase in a wave latency (electroretinogram), the foci of hepatic, myocardial and musculo-skeletal necrosis, and the decrease in white blood cells in the drug--treated/normal diet group even at the low dose of 50 mg kg-1 day-1 suggests that L1 may be unsuitable for the treatment of diseases which do not involve Fe overload. However, the low level of pathology in animals treated with iron prior to the trial suggests that even a small degree of iron overload (two-fold after eight months) is protective at this drug level. We conclude that the relationship between drug dose and iron status is critical in avoiding toxicity and must be monitored rigorously as cellular iron is depleted.