In this study, ball grid array (BGA) packages with Sn-3.0Ag-0.5Cu (SAC305) solder balls were reballed with Sn-37Pb solder balls. Three different reballing methods were used. The non-reballed lead-free BGAs were assembled with SAC305 and Sn-37Pb solder pastes to form the lead-free and mixed assemblies. The reballed Sn-Pb BGAs were assembled with Sn-37Pb solder paste to form the reballed Sn-Pb assemblies. All assemblies were subjected to a temperature cycling test with a temperature range of -55°C to 125°C. For the same component type, the reballed BGA assemblies showed similar temperature cycling reliability regardless of the reballing methods. However, the temperature cycling reliability of the reballed assemblies was worse than that of the mixed and the lead-free assemblies. The mean cycles-to-failure of the mixed assemblies was larger than or equal to that of the lead-free assemblies. Failure analysis revealed that the failure site in reballed Sn-Pb assemblies was located in the bulk solder at the component side regardless of the component type and the reballing method, indicating that the reballing method did not influence the crack propagation in reballed assemblies. The mixed assemblies had the same failure site as the lead-free assemblies, i.e., in the bulk solder at the component side. The microstructure differences between the tin-lead, lead-free, and mixed assemblies are also discussed in detail.