Dependence on benzodiazepines following continued use is by now a well-documented clinical phenomenon. Benzodiazepines differ in their dependence potential. The present studies were aimed at examining the possibility that differential rates of tolerance development might account for differences in dependence risk. Four studies are reported. The first three studies concerned normal subjects. The development of tolerance over a fifteen day period was demonstrated for three different benzodiazepines (ketazolam, lorazepam and triazolam) using two paradigms. Tolerance in terms of a reduction in effectiveness of a repeated given dose was most notable for the benzodiazepine with a medium elimination half-life (lorazepam) for physiological, behavioural and subjective measures. In the case of the drug with the longest elimination half-life (ketazolam) reduction in effectiveness could only be assumed to be occurring if account was taken of the steady increase in plasma concentrations of active metabolites. For this drug it seemed that the physiological measures were those most likely to demonstrate the development of tolerance. Although triazolam showed few significant drug effects on this paradigm (testing being 12 hours after ingestion of this short half-life benzodiazepine), tolerance was seen to develop on some subjective measures. Using an alternative method of testing tolerance, assessing responses to a diazepam challenge dose, a high degree of tolerance on two-thirds of the measures was observed in subjects when pretreated with the benzodiazepine with the most marked accumulation of active metabolites (ketazolam). The other two drugs also led to tolerance development on a range of measures; this was more marked for lorazepam than triazolam. Blunting of the growth hormone response to diazepam was the most sensitive and reliable method of detecting tolerance to the benzodiazepines. Symptoms on discontinuation of the two weeks' intake of the benzodiazepines were marked for all the drugs but unrelated to either the tolerance induced or the elimination half-life of the particular drug. A further clinical study revealed that tolerance persisted in a group of long-term benzodiazepine users for between four months and two years following complete abstinence from the drug. These patients appeared to be less affected by diazepam in terms of its commonly observed subjective effects, regardless of their original medication. These ex-long-term users of benzodiazepines were, however, more likely to manifest two specific types of effects--immediate 'symptom' reduction and exacerbation of 'withdrawal symptoms' over the subsequent week.(ABSTRACT TRUNCATED AT 400 WORDS)