Abstract: To investigate the characteristics of transverse force in the top slabs of a single-box triple-cell composite box girder with corrugated steel webs (CBGCSW) and the effective distribution width under wheel loads, a finite element model of a single-box CBGCSW with single-cell, double-cell and triple-cell was established. The regulation of transverse stress and effective distribution width of three different box girders were compared. The results of finite element analysis, which had been verified by scaled-models tests, show that the transverse force in a single-box multi-cell CBGCSW can be approximately simplified as that of a single-box single-cell box girder. However, there is a big error in calculating the effective distribution width of CBGCSW according to the current code. Using finite element models, the influences of the transverse location of wheel load, the distance between webs, the length of flange slabs, the thickness of top slabs, the size and type of corrugated steel webs on the effective distribution width of a single-box single-cell CBGCSW were analyzed. It is shown that the transverse location of wheel load and the distance between webs of loading chambers are the most important factors. The practical calculation formulas of the effective distribution width of a single-box single-cell CBGCSW were obtained by the surface fitting according to results of parametric analysis. The formulas of the effective distribution width of a single-box multi-cell CBGCSW were proposed to be 0.9 times of that of a single-cell CBGCSW. Finally, the transverse stress in the top slab of a single-box triple-cell CBGCSW were analyzed as an illustration under transverse single and multiple wheel loads, separately. The transverse stress was calculated by the elastic frame method, using the effective distribution width calculated by formulas proposed and the current code method. The results, compared with the finite element analysis, show that when transverse multiple wheel loads were imposed the calculated transverse stress may be unsafe, on the condition that the effective distribution width of different chambers were adopted as the same value. Consequently, the stress reduction factor of transverse stress in adjacent chamber which is presented in this paper are proposed to improve accuracy. The formulas presented are more accurate than those of current code, and the errors can be reduced by 20%-40%, comparing with current code method.