Long-term (mean, 8 years) users of benzodiazepines (BZs) were administered a small battery of cognitive tests on three occasions (before BZ taper, and 5 and 12 weeks post taper) as part of their BZ discontinuation program. Ninety-six patients had 5-week and 77 patients had 12-week data. For taper successes, BZ-free status was confirmed by weekly BZ plasma level determinations. Age and education, as well as baseline test scores, were used as covariates for all data analyses. Patients who successfully tapered off BZ were able to complete symbol copying (SC) and digit symbol substitution (DSST) tasks faster than patients still taking BZ (p < 0.05). In addition, using an adjective check list, patients with taper success, i.e., BZ-free patients, reported lower levels of mental and physical sedation and higher levels of tranquilization (p < 0.05) than did patients still taking BZ. These results were confirmed in two multiple regression analyses with SC and DSST as the dependent variables. Higher baseline, younger age, and successful taper status (off BZ) were selected as significant independent predictors of SC and DSST scores. In summary, cognitive functions improved for many long-term BZ users after discontinuing their BZ intake.