Aim The angiotensin‐converting enzyme gene (ACE) insertion (I)/deletion (D) polymorphism is considered a biologically plausible gene for Alzheimer's disease (AD) in cross‐sectional studies. The present study aimed to investigate the longitudinal effect of ACE I/D polymorphism on AD progression. Methods This 3‐year observational study investigated the longitudinal effect of ACE I/D polymorphism on AD progression. Clinically diagnosed AD patients with a clinical dementia rating (CDR) of 0.5 or 1 were enrolled in the study. The Mini‐Mental State Examination (MMSE), Cognitive Assessment Screening Instrument (CASI) and the CDR scale were carried out for all patients on the date of the initial interview and 36 ± 6 months after the initial evaluation. Results A total of 177 patients with sporadic AD were enrolled in this study. Among all patients, those with the I/I genotype showed a higher risk of CDR deterioration (I/I versus I/D + D/D: adjusted OR 2.103, 95% CI 1.113–3.972; adjusted P = 0.022). Among 74 AD patients without hypertension, those with the I/I genotype showed significantly greater differences in the MMSE, CASI and the CDR‐sum of box scores, and a higher risk of CDR deterioration (I/I versus I/D + D/D: adjusted OR 3.255, 95% CI 1.099–9.639; adjusted P = 0.033) after adjustment for possible confounders during the 3‐year follow up. Conclusions Patients with AD who were homozygous for the I allele presented with a more rapid AD deterioration than did those who had other ACE genotypes, particularly those patients without hypertension. Geriatr Gerontol Int 2016; ••: ••–••.