To simulate a trunnion-hub-girder (THG) assembly for bascule bridges, two full-scale laboratory tests were conducted for quantifying stresses at previously observed failure locations and for identifying a favorable assembly procedure. One assembly procedure, AP#1, cools the trunnion for a shrink fit into the hub, followed by cooling of the trunnion-hub assembly to shrink fit it into the girder. Using AP#1, development of cracks on the hub was observed in one THG assembly, and, in yet another assembly, the trunnion got stuck in the hub before full insertion could take place. Large hoop stresses and low temperatures were observed at the trunnion-hub interface when the trunnion-hub assembly was cooled for insertion into the girder. Since fracture toughness of THG parts decreases with temperature, allowable crack lengths were small. In an alternative assembly procedure, AP#2, where the hub is shrink fitted into the girder first, followed by cooling the trunnion and shrink fitting it into the hub-girder assembly, the allowable crack length was determined to be double the allowable crack length of AP#1. Hence, for the given full-scale geometry and interference values, assembly procedure AP#2 was found to be better than AP#1.