Although high blood pressure (BP) and BP variability have been reported to be associated with cognitive impairment, few studies have investigated the association between home BP (HBP) and cognitive function in the oldest-old. The aim of this study was to evaluate whether the value of and the day-to-day variability in HBP was associated with cognitive function in a Japanese community-dwelling oldest-old population. Among 111 participants aged 85–87 years, cognitive function was assessed using the Japanese version of the Montreal Cognitive Assessment (MoCA-J). HBP was measured two times every morning for a median of 30 days. The value of and variability in HBP were calculated as the average and coefficient of variation (CV) of the measurements, respectively. The associations of HBP variability with MoCA-J were examined using multiple linear regression models. Of 111 participants, 47.7% were men, and 64.0% were taking medications for hypertension. The mean HBP was 141.9 ± 14.8/72.2 ± 8.4 mmHg, and the mean CV of HBP was 6.7 ± 1.9/6.8 ± 2.4. The mean total MoCA-J score was 22.9 ± 3.5. The MoCA-J score was significantly lower with increasing CVs of both systolic BP (b = −0.36, p = 0.034) and diastolic BP (b = −0.26, p = 0.046) after adjustment for possible confounding factors. The value of HBP was not associated with MoCA-J. In the community-dwelling oldest-old population, higher day-to-day HBP variability, but not the value of HBP, was associated with cognitive impairment. When measuring HBP, attention should be paid not only to the values but also to their variations.