Our objective was to determine the prevalence and predictors of cognitive impairment in ALS, measure differences in survival among impaired and unimpaired patients, and assess changes in neuropsychological test performance over time. Fifty patients were enrolled in a prospective cohort study of neuropsychological performance. ANOVA and chi(2) tests assessed differences in clinical characteristics and neuropsychologic test results; general estimating equations assessed change in test performance; multiple regression determined which variables contributed to cognitive status; and Cox models compared survival. Thirty-six patients were categorized as cognitively normal, and 14 were impaired. Impaired patients were older at testing (p = 0.024), but no more likely to have bulbar signs. Predicators of impairment were symptom duration (p < 0.001), motor function (p < 0.001), and rate of ALS progression (p < 0.001). The Benton recognition (p < 0.001), Boston naming (p = 0.001), Wisconsin Card Sort (p = 0.001) and word generation (p = 0.001) tests contributed most strongly to cognitive status. Survival was worse in impaired patients (p = 0.027). Over time, only animal word generation declined (p = 0.016). In conclusion, 28% percent of patients were cognitively impaired. Older age and more severe ALS were associated with impairment. The strongest neuropsychological predictors of cognitive status were measures of executive, episodic memory and language function. Cognitively impaired patients had shorter survival time.