We examined the frequency of the T allele of the C270T polymorphism of the brain-derived nerve growth factor (BDNF) gene in a test and replication test design. Our objective was to determine if there is an association between the BDNF gene and Alzheimer's Disease (AD) in a US population. There were 106 autopsy-proven AD cases and 101 controls of similar ages in each test for a total of 212 AD cases and 202 controls. We found that there was a significant increase in the T allele in both the initial set (p=.04) and in the replication set (p=.018). For both groups combined p=.0008. Odds ratio=3.28, 95% CI=1.69–6.34. There were 54 cases of early-onset AD (EOAD) and 159 cases of late-onset AD (LOAD). The results were only significant for LOAD, p=.0002, odds ratio=3.81, 95% CI=1.93–7.52. The r² or fraction of the variance attributed to the BDNF gene for the LOAD cases was .046. The results were independent of the APOE ɛ4 allele. When the younger controls were removed, providing a close age match to the AD subjects, the frequency of the T allele was even lower and the differences were still significant for both total AD and LOAD cases. In a logistic regression analysis including APOE, age, sex and BDNF, BDNF was significant at p<.0001. We concluded that BDNF gene variants are significant risk factors for late onset AD.