Experiments were conducted to examine whether performance in hippocampally-mediated learning tasks is influenced by testosterone (T) and/or its 5alpha-reduced metabolites, dihydrotestosterone (DHT) and 3alpha-androstanediol (3alpha-diol). Performance in the conditioned fear and inhibitory avoidance tasks were examined in intact and gonadectomized (GDX), androgen-replaced rats. In Experiment 1, the behavior of intact and GDX rats in the conditioned fear paradigm were compared. GDX rats spent more time freezing, an index of increased learning, in the context, hippocampally-mediated task, but not in the cued, amygdala-mediated task. In Experiment 2, GDX rats were administered T, DHT, 3alpha-diol, estrogen (E2), or vehicle 1 mg/kg sc after training in the conditioned fear paradigm. T-, 3alpha-diol-, or E2-, compared with vehicle-administered rats, spent significantly more time freezing in the contextual, but not the cued, condition. In Experiment 3, intact compared with GDX rats had significantly longer crossover latencies, indicating better performance, in the inhibitory avoidance task. In Experiment 4, T, DHT, 3alpha-diol, or vehicle 1 mg/kg sc was administered to GDX rats immediately following training in the inhibitory avoidance task. Rats administered T, DHT, or 3alpha-diol had significantly longer crossover latencies compared with vehicle controls. In Experiment 5, androgen levels in the hippocampus were elevated 1 h following administration, when androgen exposure is essential for consolidation. These data indicate that androgens effects to enhance learning may be mediated in part by actions of 5alpha-reduced metabolites in the hippocampus.