Visitor questionnaire survey is a common method which recreation researchers would like to use in order to investigate visitors' states of minds toward leisure and recreation. Whether or not the questions asked in the survey are complete would affect the study results and, consequently, affect the decisionmaking processes. The purpose of this paper was to use two examples: first, definition of attitude and its relations to behavior; secondly, tests of two hypotheses underlying the Recreation Opportunity Spectrum (R.O.S.) to investigate the content validity of variable selection and construct validity of relationships between different variables, hopefully, for the references of recreation researchers in questionnaire design. This paper suggests that variables be selected on the basis of study theme and be thoroughly, clearly, and tangibly defined. In addition, the test of assumptions should completely be based on the original model or conceptual framework. By doing so, both content validity and construct validity can be enhanced.