The pharmacokinetics and plasma binding of diazepam were compared in man, dog, rabbit, guinea pig and rat. Diazepam (D) and its major metabolite, desmethyldiazepam, were measured in blood and plasma by a specific and sensitive GLC assay with an electron capture detector. After an i.v. bolus injection, plasma levels of D declined biexponentially in all species examined and the data were analyzed according to the 2 compartment open model. The binding of D and desmethyldiazepam was determined at therapeutic concentrations by equilibrium dialysis in man (96.8 and 96.6%, respectively), dog (96.0 and 95.9%), rabbit (89.9 and 94.7%), guinea pig (91.3 and 78.6%) and rat (86.3 and 90.5%). In man, the elimination half-life, T1/2(.beta.), increased significantly (P < .01) with decreasing total plasma clearance .**GRAPHIC**. Plasma binding affected Vd [vol of distribution] and .**GRAPHIC**. but only .**GRAPHIC**. increased significantly (P < .05), if more free D was available. Unbound drug is rate-determining for clearance by the liver, and that D fits into the restrictive elimination class in man. In the 4 animal species tested, .**GRAPHIC**. was a direct linear function of the body surface area. T1/2(.beta.) and the rates of drug clearance were characteristic figures for each species: from 1.1 h and 81.6 ml/min per kg in the rat to 32.9 h and 0.35 ml/min per kg in man; T1/2(.alpha.), the half-life of distribution, varied only approximately 3-fold (0.3-1.0 h) in the different species. A considerably higher extraction ratio than the unbound fraction of diazepam exists in these animal species, and blood clearance exceeds liver blood flow, giving reason to assume a much higher ability of the liver to metabolize D, and a species-dependent extrahepatic metabolism. The large variations described suggest that pharmacokinetic data or plasma binding results cannot simply be extrapolated to man.