In order to investigate whether alcohol-withdrawal kindling is an irreversible process, male Wistar rats were exposed to 12 episodes, each consisting of 2 days of severe alcohol intoxication and 5 days of alcohol withdrawal. Spontaneous withdrawal seizures were found in 15% of the animals during episodes 10-12. After an alcohol-free period of 26 days, the animals were subjected to three more episodes of alcohol dependence (i.e. episodes 13-15) in which 12% of the animals developed spontaneous withdrawal seizures. Based on several statistical tests, we concluded that there was no true difference between the seizure activity in episodes 10-12 and episodes 13-15, indicating that alcohol-withdrawal kindling is a long-lasting and perhaps irreversible process. In a second experiment, an alcohol-withdrawal kindled group was first exposed to seven episodes of alcohol dependence. A diazepam group went through the same alcohol regimen, but each withdrawal reaction was blocked by diazepam treatment. Finally, a single episode group was included which was fed isocalorically with the kindled animals. After an alcohol-free period of 11 days, all three groups were subjected to 4 days of severe alcohol intoxication. During the subsequent withdrawal reaction seizures were observed in 22-26% of the animals with no significant differences across the groups. These results call for a modification of the kindling hypothesis of alcohol withdrawal and suggest that kindling-induced alterations may be overlooked if convulsive behaviour is tested during a relatively strong withdrawal reaction.