Objective: This study examined the relation between two risks for Alzheimer’s disease (AD): the apolipoprotein (APOE) Ε4 allele and poor memory test performance. Methods: In the Canadian Study of Health and Aging (CSHA), a 5-year longitudinal population-based study that screened and followed over 10,000 participants, 2,914 had an initial clinical assessment and 1,624 had APOE genotype testing. All participants were categorized as having no cognitive impairment, cognitive impairment but no dementia, or dementia at both baseline and follow-up. We examined those (n = 209) with a complete neuropsychological assessment at baseline and no evidence of cognitive impairment who had either APOE Ε3/Ε3 or Ε3/Ε4 genotypes and who had a clinical consensus diagnosis of either no cognitive impairment or AD at follow-up. Delayed free recall memory was evaluated at CSHA-1 with the Buschke Cued Recall Test (BCRT). Results: The risk of AD at follow-up was increased for participants with an APOE Ε3/Ε4 genotype when memory test performance was not considered, but logistic regression demonstrated that a model which also considered baseline memory test performance was more predictive of AD. In the more complete model, reduced BCRT free recall scores were associated with an increased risk of AD, whereas the risk associated with the APOE Ε3/Ε4 genotype was no longer significant. Conclusions: For those with no evidence of cognitive impairment, drawn from a population-based sample of elderly persons, the APOE Ε3/Ε4 genotype was only associated with an increased risk of AD after 5 years if their memory test performance was relatively poor at baseline. Regardless of the APOE genotype, and in the absence of clinical evidence of cognitive impairment, reduced scores on a test of delayed free recall at baseline was associated with an increased risk of AD after 5 years.