It has often been observed that periods of abuse of non-narcotic drugs precede the development of dependence on opiates, but this correlation does not necessarily imply that there is a simple causal relationship. We have previously described a procedure for inducing rats to drink solutions of morphine in preference to water; self-administration of the drug in this way seems to be a valid model of dependence. For the first 46 days of the experiment described here, solutions of alcohol, amylobarbitone, chlordiazepoxide, cocaine or dexamphetamine were made available to different groups of rats. Most of these drugs were ingested in substantial doses, had clear effects on behaviour and produced characteristic patterns of drinking over the repeated trials. However, there was no unequivocal evidence of dependence, and indeed the rats learned to reject the solutions of dcxamphetamine. Ingestion of these drugs did not affect the eventual development of dependence when solutions of morphine were substituted at a later stage, although the avoidance of dexamphetamine seemed to temporarily transfer to morphine. Further studies using other methods and species are needed before inferences can be made about "escalation" to dependence on opioids in man.