The behavioral and neurohistological protective effects of tacrolimus (FK506) were examined in rats subjected to 15-min global forebrain ischemia. Learning and memory performance were evaluated in an aversive, non-food-motivated, eight-arm radial maze. In one experiment, naive rats were rendered ischemic, and 15 days later they were tested for acquisition of a spatial task (postoperative training). In a complementary experiment, rats were trained for 8 days and then subjected to ischemia (preoperative training); 15 days later (on Day 24 of testing) they were retested for retention of cognition. FK506 (1.0 mg/kg) was given intravenously at the beginning of reperfusion, followed by doses applied intraperitoneally 6, 24, 48 and 72 h postischemia. Behavioral performance was expressed by latency to find the goal box, and number of errors. Ischemia did not affect acquisition performance. In contrast, retention of cognition was markedly impaired by ischemia, particularly working memory (P<.05-.001). This ischemia-induced, retrograde amnesia was significantly reduced by FK506 compared to vehicle alone on Day 24, as measured by latency and working memory errors (P<.025). A neuroprotective effect of FK506 was also seen on working memory, when postischemic performance was compared to that prior to ischemia (P>.05, Day 24 vs. Day 8, paired samples), in contrast to the significant, retrograde amnesia found in the ischemic, vehicle-treated group (P<.01). FK506 also significantly reduced the extent of hippocampal CA1 cell loss; however, this effect did not correlate with behavior. The present results suggest that the histological, neuroprotective effect of FK506 may be accompanied by a reduction in cognitive impairment, as assessed in a novel, non-food-motivated, eight-arm radial maze after transient, global, cerebral ischemia in rats.