We sought to longitudinally evaluate the potential association of educational level with performance on verbal and nonverbal tasks in individuals with mild cognitive impairment (MCI). We evaluated patients with MCI, age >50 years, no medication intake, absent vascular risk factors, and no lesions on brain magnetic resonance imaging (MRI). Each patient underwent a clinical assessment packet and a series of neuropsychological tests of the language and constructional praxis subtests of Cambridge Cognitive Examination (CAMGOG) and the Boston naming test (BNT), at baseline, 6 months, and 12 months. Educational levels were defined taking into account the total years of education, the school level, and diplomas. MCI patients with low education level showed a stepwise reduction in scores of naming objects (NO; P = 0.009), definition (DF; P = 0.012), language (LT; P = 0.021), constructional praxis (CD; P = 0.022), confrontation naming skills (BXB; P = 0.033), phonemic help (BFB; P = 0.041), and BNT (P = 0.002). Analysis of covariance, controlling for baseline scores, showed that education was associated with NO score (P = 0.002), DF score (P = 0.005), LT (P = 0.008), CD score (P = 0.008), BXB score (44.36 ± 1.84, P = 0.0001), BFB (P = 0.022), and BNT (P = 0.004). Our findings indicate that education appeared to affect verbal and nonverbal task performance in MCI patients. Despite the fact that higher educated patients are more acquainted with the tasks, slower deterioration in consecutive follow-up examinations could be explained by the cognitive reserve theory. The potential association of this protective effect with delayed onset of symptoms deserves further investigation.