Six monkeys were trained preoperatively in an automated object-in-place memory task in which they learned 20 new scenes in each daily session. Three of the six monkeys then received stereotaxically guided bilateral mamillary body lesions, leaving the fornix intact, while the other three received a control operation. Postoperatively the control animals' rate of learning new scenes was unchanged, but the animals with mamillary body lesions showed a severe impairment, equal to that seen in previous experiments after fornix transection. All six animals were then given fornix transection, in addition to the existing mamillary or control operation. The control group now showed, after fornix transection, an impairment equal to that of the animals with mamillary body lesions alone. But the animals with mamillary body lesions did not show any additional impairment following fornix transection. We conclude that (1) the role of the mamillary bodies in a model of human episodic memory is as important as the role of the fornix, (2) the fornix and mamillary bodies form a single functional memory system, since the effect of lesions in both parts is no more severe than the effects of a lesion in one of the parts alone, and (3) the idea that the functional effects of fornix transection result from cholmergic deafferentation of the hippocampus receives no support from the present results; rather, they support the idea that in primates the fornix and mamillary bodies, together with connected structures, including the subiculum, mamillo-thalamic tract, anterior thalamic nuclei, and cingulate bundle, form a cortico-cortical association pathway for episodic memory.