Factors influencing the rate and extent of benzodiazepine uptake into CSF, peripheral tissue distribution and EEG effects were evaluated in a model using anesthetized male cats. A single (0.25-10 mg/kg) dose of the following 8 benzodiazepines was administered i.v.: diazepam, demethyldiazepam, midazolam, lorazepam, alprazolam, triazolam, flunitrazepam and clobazam. Multiple samples were simultaneously drawn from arterial blood and cisternal CSF over the next 4 h and the EEG was continuously monitored. Concentrations of benzodiazepines in plasma and CSF samples were measured by electron-capture GLC and plasma protein binding determined by equilibrium dialysis. Physicochemical properties of lipophilicity of each benzodiazepine were determined by measurement of the octanol/buffer partition ratio at physiologic pH and by the high-pressure liquid chromatographic (HPLC) retention on a reverse-phase C18 column at neutral pH. Disappearance of all benzodiazepines from plasma was consistent with a linear sum of 2 or 3 exponential terms. After correction for individual differences in protein binding, volume of distribution (Vd) of unbound drug was highly correlated with HPLC retention (r = 0.91), but not significantly related to octanol/buffer partition coefficient. Diazepam and midazolam, having the longest HPLC retention also had the largest unbound Vd. All benzodiazepines rapidly entered CSF, with peak concentrations usually attained within 15 min of dosage. More lipophilic drugs tended to enter CSF most rapidly, but associations of entry rate and in vitro lipophilicity were not significant. After distribution equilibrium was attained, disappearance of benzodiazepines from both plasma and CSF occurred in parallel. Equilibrium CSF/total plasma concentration ratios of all drugs were much less than unity. This was explained by plasma binding because equilibrium ratios were highly correlated with free fraction (r = 0.93, regression line slope = 0.98). All benzodiazepines produced diffuse low frequency (1-5 Hz) EEG activity of very rapid onset after the i.v. dose. Onset rate was not correlated with in vivo lipophilicity. The duration of slow-wave activity differed markedly among drugs, having a high negative correlation with log of HPLC retention (r = 0.77) and with log of in vivo unbound Vd (r = 0.87). Benzodiazepine distribution in vivo is determined largely by lipid solubility, which in turn is reflected best by retention on a reverse-phase HPLC system. The rate of CSF entry, as well as the rate of onset of EEG slow-wave effects, were rapid and not significantly related to lipophilicity, but duration of EEG effects were longest for the least lipophilic drugs having the smallest Vd.