A cross-over study was performed in 6 healthy mixed-breed dogs and 4 healthy Beagles. Diazepam was administered per rectum to Beagles (0.5 mg/kg of body weight) and mixed-breed dogs (2 mg/kg), and IV (0.5 mg/kg) to both groups of dogs. Each dog received the drug by both routes, with a 1-week washout period between dosages. After diazepam administration, blood samples were collected to measure plasma concentration of diazepam and its active metabolites, desmethyldiazepam and oxazepam, by use of reverse-phase high-performance liquid chromatography (HPLC). Systemic availability was assessed by comparing the area under the curve for diazepam metabolites for each route of administration. Mean (+/- SD) diazepam concentrations in plasma after rectal administration were low in comparison with those obtained after IV administration, with systemic availability of only 7.4 (+/- 5.9) and 2.7 (+/- 3.2)% for the high and low dose, respectively. However, diazepam was converted to its metabolites within minutes after administration. Accounting for the total concentration of benzodiazepines (diazepam plus desmethyldiazepam and oxazepam) in plasma, systemic availability was 79.9 (+/- 20.7) and 66.0 (+/- 23.8)% for the high and low dosage, respectively. After IV administration, diazepam concentration decreased, with a half-life of only 14 to 16 minutes, but desmethyldiazepam and oxazepam concentrations decreased more slowly, with a half-life of 2.2 to 2.8 hours and 3.5 to 5.1 hours, respectively. Each of the metabolites is reported to have anticonvulsant activity. After rectal administration of the high dose, mean total benzodiazepine concentration was above 1.0 microgram/ml within 10 minutes and was maintained above this concentration for at least 6 hours. We conclude that diazepam is absorbed after rectal administration in dogs, and that the pharmacologic effects are probably caused by the active metabolites, not the parent drug. Samples also were analyzed by use of a nonspecific commercial benzodiazepine fluorescence polarization immunoassay (FPIA). Correlation between the FPIA and HPLC assay was strongest for diazepam (R2 = 0.84), weak for desmethyldiazepam (R2 = 0.09), and nonexistent for oxazepam. We conclude from a comparison of assays that HPLC is preferred over the FPIA method for measuring benzodiazepines in dogs.