Although emphasis on educational value, which leads to underestimation of risk, is pointed out as one of the factors contributing to incidents in school, this has not been demonstrated. The purpose of the present study was to compare between elementary school teachers and college students attending a teacher training course to assess how emphasis on educational value affects risk perception for physical education activities in elementary schools, and to determine how perception of risk by elementary school teachers affects its estimation. Respondents (75 elementary school teachers and 61 college students attending a teacher training course; total 136) were requested to evaluate 12 physical education activities using 6 rating scales of risk perception (Voluntariness, Knowledge about risk, Control over risk, Common-dread, Severity of consequences, and Chronic-catastrophic) and 3 educational values (Touching, Educational value, and Importance as school activity). As a result of three-mode factor analysis (Tucker, 1966), 3 factors were extracted from both the rating scale mode (Dread, Controllability, Educational value) and the target mode (Overcome activities, Fixed-form activities, Contact activities), and 5 factors were extracted from the individual mode. The core array, which indicates individual tendency, revealed that Educational value was correlated with Dread only in Overcome activities (swimming, endurance running). This was interpreted to indicate that an educational value different from general educational values may be perceived in the Overcome activities. The assessment of educational values differed between the elementary school teachers and the college students in terms of Dread for Contact activities, and in Educational value for all activities and also for Overcome activities. In addition, multiple regression analysis revealed that the interaction effect of Dread and Educational value on elementary school teachers’ risk estimation (the number of fracture accidents per year) was evident only in Overcome activities, which meant that educational value affects risk estimation as a moderating factor. These results indicate that risk estimation for physical education activities becomes more precise with practical experience in education, although a cognitive bias due to Dread and Educational value exists in Overcome activities.