This paper represents one portion of an ongoing effort to elaborate on pedagogical principles based on structural-developmental (largely Piagetian) theory and research. Here, a concept of "problem structure" is developed as a means for identifying physical factors that contribute to children's ability to apply experimental reasoning. Data are presented that demonstrate change in stage of reasoning as a function of change in problem structure: in this case, "mixed" versus "well-separated" variables. It is proposed that level of reasoning results from an interaction between cognitive structure and problem structure. The issue of horizontal decalage (developmental lags) is explicated in these terms. Principles for structuring physical problems for teaching are presented. (RH)