A total of 1207 elderly people were selected by multistage probability sampling (number of children of the elderly = 201). We used a self-designed behavior knowledge questionnaire (38 questions) to survey the participants. After the 38 questions were answered, the total score was determined. The group with adequate health behavior knowledge (score ≥ 24) and that with inadequate knowledge (score < 24) were distinguished by logistic regression to explain the influencing factors affecting health knowledge.