Objective: To develop a quantitative method of assessing cognitive anosognosia in six cognitive and two noncognitive domains. Methods: Control (n = 32) and probable Alzheimer disease (pAD) (n = 14) subjects self-estimated memory, attention, generative behavior, naming, visuospatial skill, limb praxis, mood, and uncorrected vision, both before and after these abilities were assessed. Based on this estimate and their performance the authors calculated an anosognosia ratio (AR) by dividing the difference between estimated and actual performance by an estimated and actual performance sum. With perfect awareness, AR = 0. Overestimating abilities would yield a positive AR (≤1); underestimation would yield a negative AR (≥-1). Results: Relative to controls, pAD subjects demonstrated anosognosia. Pre-testing (off-line), pAD subjects overestimated their visuospatial skill; post-testing (on-line), pAD subjects overestimated their memory. Control subjects also made self-rating errors, underestimating their attention pre-testing and overestimating limb praxis and vision post-testing. Conclusions: This anosognosia assessment method may allow more detailed examination of distorted self-awareness. These results suggest that screening for anosognosia in probable Alzheimer disease (pAD) should include self-estimates of visuospatial function, and that, in pAD, it may be useful to assess anosognosia for amnesia both before and after memory testing.