Top-down proteomics using mass spectrometry facilitates the identification of intact proteoforms, that is, all molecular forms of proteins. Multiple past advances have lead to the development of numerous sample preparation workflows. Here we systematically investigated the influence of different sample preparation steps on proteoform and protein identifications, including cell lysis, reduction and alkylation, proteoform enrichment, purification and fractionation. We found that all steps in sample preparation influence the subset of proteoforms identified (for example, their number, confidence, physicochemical properties and artificially generated modifications). The various sample preparation strategies resulted in complementary identifications, substantially increasing the proteome coverage. Overall, we identified 13,975 proteoforms from 2,720 proteins of human Caco-2 cells. The results presented can serve as suggestions for designing and adapting top-down proteomics sample preparation strategies to particular research questions. Moreover, we expect that the sampling bias and modifications identified at the intact protein level will also be useful in improving bottom-up proteomics approaches.