Despite manufacturing sub-components to a high precision, large over-constrained assemblies are often impossible to assemble to tolerance limits when variations are present. This necessitates expensive and time consuming variation management processes at assembly, such as shimming. Existing research has not established a methodology to model the variation propagation mechanisms that results in this assembly variation. This paper presents such a methodology, which has the ability to quantify the assembly variation of over-constrained assemblies at the planning stage, providing useful data for determining the most appropriate combination of fabrication and assembly processes to use for a given case. The methodology is validated using an aerospace wing spar assembly, and a sensitivity study completed to rank the key variation drivers in the over-constrained assembly.