I n rats, the effects of an intracerebroventricular (ICV) nicotinic agonist nicotine (NIC), the nicotinic antagonist mecamylamine (MEC), and combinations of NIC + MEC were assessed in a radial-arm maze (RAM). In experiment 1, exploratory behavior was assessed in untrained rats (N = 13). The rats received 4 kg, 8.65 nmol NIC (NIC4), 200 kg, 0.98 pmol MEC (MEC), and saline (SAL) ICV infusions. NIC 4 caused a significant increase in choice distribution compared to SAL ( P < 0.025). In experiment 2, rats (N = 10) were trained to perform a working memory task for food reinforcement in the RAM. ICVdoses of SAL, NIC4, NIC8, MEC, MECNIC4, and MECNIC 8 were administered after completion of the training period. MEC caused a significant deficit in choice accuracy when compared to SAL ( P < 0.025). This deficit was reversed when NIC 8 was coadministered with MEC ( P < 0.05). There were no significant effects on choice latency for either study. The effects of ICV NIC and MEC on RAM performance are generally similar to their systemic effects in that NIC improves and MEC impairs choice accuracy. The reversal of the MEC-induced choice deficit by ICV NIC administration has not been reported with systemic administration. ICV MEC induces a choice accuracy deficit without increasing choice latency. This has not been seen with systemic MEC administration. The current result implies that the MEC-induced choice accuracy deficit did not result from MEC-induced sedation. The data indicate that previously reported changes in choice accuracy from peripherally administered NIC and MEC result from their central effects.