Limitations of preclinical models of human memory contribute to the pervasive view that rodent models do not adequately predict therapeutic efficacy in producing cognitive impairments or improvements in humans. We used a source-memory model (i.e. a representation of the origin of information) we developed for use in rats to evaluate possible drug-induced impairments of both spatial memory and higher order memory functions in the same task. Memory impairment represents a major barrier to use of NMDAR antagonists as pharmacotherapies. The scaffolding protein postsynaptic density 95kDa (PSD95) links NMDARs to the neuronal enzyme nitric oxide synthase (nNOS), which catalyzes production of the signaling molecule nitric oxide (NO). Therefore, interrupting PSD95-nNOS protein-protein interactions downstream of NMDARs represents a novel therapeutic strategy to interrupt NMDAR-dependent NO signaling while bypassing unwanted side effects of NMDAR antagonists. We hypothesized that the NMDAR antagonist MK-801 would impair source memory. We also hypothesized that PSD95-nNOS inhibitors (IC87201 and ZL006) would lack the profile of cognitive impairment associated with global NMDAR antagonists. IC87201 and ZL006 suppressed NMDA-stimulated formation of cGMP, a marker of NO production, in cultured hippocampal neurons. MK-801, at doses that did not impair motor function, impaired source memory under conditions in which spatial memory was spared. Thus, source memory was more vulnerable than spatial memory to impairment. By contrast, PSD95-nNOS inhibitors, IC87201 and ZL006, administered at doses that are behaviorally effective in rats, spared source memory, spatial memory, and motor function. Thus, PSD95-nNOS inhibitors are likely to exhibit favorable therapeutic ratios compared to NMDAR antagonists.