The long-span culvert is a synergistic unit composed of a corrugated metal liner and a compacted soil envelope that surrounds the liner. Conceptually, the system is very simple and, therefore, economically attractive as a bridge substitute. Analytically, however, the system is not simple because of the modeling difficulties associated with soil-structure interaction. Using the finite element method, this study investigates the influence of fundamental modeling assumptions on the behavior of long-span culverts. Two basic modeling assumptions are examined: large deformation theory versus small deformation theory and monolith structure versus incremented structure. In addition, the sensitivity of the following parameters are determined: compaction loads, soil stiffness, liner gage, liner shape, and special features of manufacturers. Results are shown graphically by comparing crown displacement histories between parametric families. Comparisons of maximum moment and thrust are also reported. Based on these studies, recommendations for analytical modeling techniques are summarized. The intent of this study is to provide a foundation for other studies. A systematic investigation of modeling assumptions and parameter sensitivity is a necessary step toward an analytical model for long-span culverts./Author/