The Trail Making Test (TMT) is a common two-part neuropsychological test, in which visuospatial ability (TMT-A) and executive function (TMT-B) are evaluated. Normative data for this test have not been reported for Japanese subjects; therefore, the purpose of the present paper was to investigate the effect of age and education on the TMT in 155 healthy elderly adults with clinical dementia rating 0 (healthy). The participants were classified into three groups based on age (70–74 years, 75–84 years and ≥85 years), and also into three groups based on educational level (6 years, 8 years and ≥10 years). The time to complete TMT-A and TMT-B were measured, and the difference in score between TMT-A and TMT-B (B–A) and the ratio of the score (B/A) were calculated as indices of executive function. The time for completion of both parts of the TMT increased markedly in the ≥85-years group. For TMT-A, there was a significant difference between the 6-years and 8-years groups, and between the 6-years and ≥10-years groups, and for TMT-B, there was a significant difference between the 6-years and ≥10-years groups, and between the 8-years and ≥10-years groups. The difference and ratio scores increased in the ≥85-years group, but the educational level did not significantly influence these scores. Our data suggest that cognitive functions evaluated by TMT-A and TMT-B are not affected by aging until the subjects are ≥85 years old. For TMT-A, an educational effect becomes apparent when the population includes poorly educated subjects, but this part of the test is not affected by educational level provided that the subjects have some education (>6 years). The time to complete TMT-B is affected by educational level, consistent with previous reports. However, when adjusted using the results for TMT-A [(B-A) or (B/A)], the educational effect on executive function disappeared. Thus, the effect of educational level on executive function was unclear in normal elderly subjects.