Three experiments examined the effects of the training-to-testing inte
rval on alcohol aversions. In Experiments 1 and 2, rats learned a tast
e aversion to a 4% (v/v) alcohol solution using lithium chloride as th
e illness agent. With a between-groups design, subjects were tested wi
th 1%, 4%, or 7% alcohol, beginning either 2 or 21 days after training
. In both experiments, results showed that rats learned aversions to t
he trained 4% alcohol that generalized to the nontrained 1% and 7% con
centrations. Furthermore, in Experiment 1 aversions to 1% and 7% alcoh
ol were stronger in groups tested 21 days after training relative to g
roups tested 2 days after training. When the strength of the illness a
gent was reduced in Experiment 2, aversions to all concentrations of a
lcohol were stronger at the delayed testing interval. Experiment 3 rul
ed out the possibility that enhanced alcohol avoidance at delayed test
ing was the result of spontaneous recovery of neophobia. The results s
uggest that taste aversions to alcohol become stronger with time. Poss
ible mechanisms for this ''incubation effect'' are discussed. The pres
ent findings have implications for improving emetic therapy as a treat
ment for human alcoholics.