Combinations of drugs approved to treat Alzheimer's disease (AD) were tested in older rabbits with delay eyeblink classical conditioning, a form of associative learning severely impaired in AD. In Experiment 1 (n ¼ 49 rabbits), low doses (0.1, 0.5, 1.0, and 0.0 (vehicle) mg/kg) of memantine (Namendat) were tested. These three doses neither improved nor impaired acquisition at a statistically significant level. The 0.5 mg/kg dose had the greatest effect numerically and did not cause sensitization or habituation in explicitly unpaired controls. In Experiment 2 (n ¼ 56), doses of galantamine (Razadynet; 3.0 mg/kg) and donepezil (Ariceptt; 0.75 mg/kg) that had comparable magnitudes of cholinesterase inhibition were tested alone and in combination with 0.5 mg/kg memantine. Older rabbits treated with galantamine and with galantamine + memantine learned significantly better than vehicle-treated rabbits, but adding memantine did not improve learning over galantamine alone. Older rabbits treated with donepezil or a combination of memantine and donepezil did not learn significantly better than rabbits treated with vehicle. Galantamine has two mechanisms of action: mild cholinesterase inhibition and allosteric modulation of nicotinic acetylcholine receptors (nAChRs). When equated for cholinesterase inhibition, galantamine had significant efficacy in the eyeblink conditioning model system, but donepezil did not, indicating that modulation of nAChRs may be the mechanism that significantly ameliorates learning deficits in this model. In the absence of AD neuropathology in older rabbits, memantine had no efficacy alone or in combination with the other drugs.