This study aimed to determine whether visit-to-visit blood pressure (BP) variability (BPV) is associated with incident frailty. We included 1 394 nonfrail community-dwelling participants aged ≥70 years from the Multidomain Alzheimer Preventive Trial (MAPT) who underwent repeated clinical examinations, including BP and frailty, over a 5-year follow-up period. Systolic BPV (SBPV), diastolic BPV (DBPV), mean arterial pressure variability (MAPV), and pulse pressure variability (PPV) were evaluated using standard deviation (SD), coefficient of variation (CV), average real variability, successive variation, variation independent of mean, and residual SD. Incident frailty was assessed using the Fried phenotype. Cox proportional hazards models were used for the analyses. Higher SBPV was significantly associated with greater risk of frailty (1-SD increase of CV: hazard ratio [HR] = 1.18, 95% confidence interval [CI]: 1.02-1.36) after adjustment for demographics, systolic BP, antihypertensive drugs, body mass index, diabetes, ischemic heart disease, congestive heart failure, stroke, atrial fibrillation, MAPT randomization group, and frailty status. Similar results were observed with all indicators of variability. Higher PPV was associated with a greater risk of developing frailty over time (1-SD increase of CV: HR = 1.17, 95% CI: 1.01-1.35). DBPV and MAPV were not significantly associated with incident frailty. Higher SBPV and PPV were associated with greater risk of incident frailty. Our findings support the concept of BP physiological dysregulation underlying the frail state and suggest that BP instability could be an early marker of frailty.