The edge drop of cold-rolled strip directly affects width accuracy and edge quality of product, it is especially prominent under the conditions of non-oriented silicon steel with wide specification and limited rolling reduction. In the paper, the integrated three-dimensional finite element model coupling with roller-strip steel is established based on the 1 780 mm 6-high CVC (continuously variable crown) rolling mill. Such aspects as geometric modeling, mesh refinement at the edge and contact zone between rollers, boundary constraint, friction contact setting and multi-step load incremental control of this model are optimized. Compared with the measured sectional contour of the unit, the RMSE (root mean square error) of simulation result is 6.85 μm and MAPE (mean absolute percentage error) is 1.26%, which indicate that the model is with higher prediction accuracy. The influences of thickness, width and reduction ratio of strip steel as well as deformation resistance of materials on the formation laws of edge drop are systematically analyzed. The transverse metal flow at the edge could be significantly amplified with the increase of thickness or reduction ratio so as to aggravate edge drop; transverse flow at unit width could be reduced by increasing width of strip steel so as to weaken edge drop to some extent; it is beneficial to inhibiting metal flow at the edge to increase deformation resistance so that edge drop and its sensitivity to reduction ratio are reduced. The simulation system of limited conditions is established based on this to investigate the comprehensive abilities of regulation and control of roll bending forces of working roll and intermediate roll as well as amount of roll shifting for intermediate roll to edge drop. The results showed that the effects of working roll bending on edge drop control were the most significant, the average control effect was approximately 3.5 times of that of intermediate roll bending and about 30 μm of linear regulation amplitude could be achieved for intermediate roll shifting at its positive and negative limiting positions. The efficiency and sensibility of different control methods are with quantitative evaluation by introducing the efficacy coefficients of edge drop control, which could provide the theoretical basis and engineering references for optimization design of roll shape for 6-high CVC rolling mills and formulating the process of edge drop control for silicon steel with wide specification.