Rats were trained on an olfactory continuous delayed nonmatching-to-sample (DNMTS) task and then given 1 of 4 treatments: sham surgery or radio-frequency lesion of the lateral internal medullary lamina of the thalamus or of the frontal cortex along the medial wall or dorsal to the rhinal sulcus. Thalamic lesions produced persistent deficits on the continuous DNMTS task, and both the cortical lesions resulted in transient impairments that disappeared with continued training. Manipulations of stimulus set size and the delay between trials affected continuous DNMTS performance but did not exacerbate group differences. All 3 lesion groups performed normally when next trained on a discrimination task with odorants and go-no-go procedures comparable to continuous DNMTS. These results indicate that lesions did not affect ability to perform go-no-go procedures, to discriminate among odorants, or to use reference memory to respond on the basis of a fixed stimulus-response rule.