Objectives. To estimate the prevalence of subjective memory complaints (SMCs) in a sample of community-dwelling, older adults and to examine cognitive bases of these complaints. Participants. 499 community-dwelling adults, 65 and older. Measurements. A telephone survey consisting of cognitive tests and clinical and sociodemographic variables. SMCs were based on subjects' evaluations and subjects' perceptions of others' evaluations. Analysis. Logistic regression was used to model the risk for SMCs as a function of the cognitive, clinical, and sociodemographic variables. We tested for interactions of the cognitive variables with age, education, and gender. Results. 27.1% reported memory complaints. Among the younger age, better objective memory performance predicted lower risk for SMCs, while among the older age, better memory had no effect on risk. Among the better-educated people, better global cognitive functioning predicted lower risk for SMCs, while among the less-educated people, better global cognitive functioning had no effect on SMC risk. When predicting others' perceptions, better objective memory was associated with lower risk for SMCs. Conclusion. Objective memory performance and global cognitive functioning are associated with lower risk for SMCs, but these relationships are the strongest for the younger age and those with more education, respectively. Age and education may affect the ability to accurately appraise cognitive functioning.