Transcranial magnetic stimulation combined with intracranial EEG (TMS-iEEG) has emerged as a powerful approach for probing the causal organization and dynamics of the human brain. Despite its promise, the presence of TMS-induced artifacts poses significant challenges for accurately characterizing and interpreting evoked neural responses. In this study, we present a practical preprocessing pipeline for single pulse TMS-iEEG data, incorporating key steps of re-referencing, filtering, artifact interpolation, and detrending. Using both real and simulated data, we systematically evaluated the effects of each step and compared alternative methodological choices. Our results demonstrate that this pipeline effectively attenuated various types of artifacts and noise, yielding cleaner signals for the subsequent analysis of intracranial TMS-evoked potentials (iTEPs). Moreover, we showed that methodological choices can substantially influence iTEPs outcomes. In particular, referencing methods might strongly affect iTEP morphology and amplitude, underscoring the importance of tailoring the referencing strategy to specific signal characteristics and research objectives. For filtering, we recommend a segment-based strategy, i.e., applying filters to data segments excluding the artifact window, to minimize distortion from abrupt TMS-related transients. Overall, this work represents an important step toward establishing a general preprocessing framework for TMS-iEEG data. We hope it encourages broader adoption and methodological development in concurrent TMS-iEEG research, ultimately advancing our understanding of brain organization and TMS mechanisms.