Two earlier studies found that rhinal cortex ablations in the monkey (Macaca fascicularis) impaired delayed matching-to-sample (DMS) when the stimuli in the experiment came from a large population of possible stimuli, but not when the stimulus population was small, while uncinate fascicle section had no effect on DMS whatever the stimulus population size. The mediodorsal thalamus receives a large projection from the rhinal cortex, and has been implicated in recognition memory performance. We trained monkeys preoperatively in delayed matching-to-sample with large and small stimulus populations, exactly as in the earlier studies, then examined the effect of bilaterally ablating the medial portion of the mediodorsal thalamic nucleus. Mediodorsal lesion impaired postoperative delayed matching-to-sample performance with a large stimulus set, but had no effect on performance of DMS with a small stimulus population. In comparison with the earlier data from rhinal cortex lesions with the same methods, wherever a deficit was seen in the rhinal-lesioned animals the mediodorsal thalamic nucleus-lesioned animals showed a smaller deficit. We conclude that other efferents from the rhinal cortex, possibly those to the adjacent inferior temporal cortex, enable better performance in the mediodorsal thalamic nucleus-lesioned animals than in the animals with rhinal cortex ablation.