The use of data-independent acquisition methods such as SWATH for mass spectrometry based proteomics is usually performed using peptide MS/MS reference ion assay libraries which enable identification and quantitation of peptide peak areas. Reference assay libraries can be generated locally through information dependent acquisition, or obtained from shared data repositories for commonly studied organisms. However, there have been no studies performed to systematically evaluate how locally-generated or repository-based assay libraries affect SWATH performance for proteomic studies. To undertake this analysis we developed a software workflow, SwathXtend, which generates extended peptide assay libraries using a local seed library and delivers statistical analysis of SWATH-based sample comparisons. We designed test samples using peptides from a yeast extract spiked into peptides from human K562 cell lysates at different ratios to simulate common protein abundance change comparisons. SWATH-MS data with 2, 5 and 10% of yeast peptides spiked into the human cell lysate were assessed using several local and repository-based assay libraries of different complexities and proteome compositions. We evaluated detection specificity and accuracy to detect differentially abundant proteins and reporting thresholds for statistical analyses. We demonstrate that extended assay libraries integrated with local seed libraries achieve better performance than local limited assay libraries alone from the aspects of the number of peptides and proteins identified and the specificity to detect differentially abundant proteins; the performance of extended assay libraries heavily depend on the similarity of the seed and add-on libraries; statistical analysis with multiple testing correction can improve the statistical rigor needed when using large, extended assay libraries.