The relationship between habitual sleep and cognition in older adults with sleep complaints is poorly understood, because research has focused on younger adults, used experimental or retrospective quasi-experimental designs, and generally produced equivocal results. Prospective studies using sleep diaries are rare, but may provide important insights into this relationship as they offer greater ecological validity and allow for examination of the impact of night-to-night variability in sleep (an often overlooked aspect of sleep) on cognitive performance. Seventy-two older adults (M(age) = 70.18 years, SD(age) = 7.09 years) completed fourteen consecutive days of sleep diaries and paper/pencil self-administered cognitive tasks, including measures of processing speed (Symbol Digit) and reasoning (Letter Series). Regression analyses revealed increased average total wake time (TWT) during the night was associated with higher Symbol Digit scores, β = 0.45, P < 0.05. Night-to-night variability in either total sleep time (TST) or TWT was not associated with either cognitive measure. Implications and potential explanations for these initially counterintuitive findings are discussed.