Study of non-state violent actors can be impeded by a variety of challenges inherent to the these organizations. Methodologies not well-suited to these dynamics may result in peril for the researcher and/or her subjects, and lead to perhaps less rigorous findings. We argue here that the ontology of process tracing is conducive to highly rigorous and safe examination of NSVAs. Here, we highlight a collection of data collection strategies that are particularly useful in this research; introduce and evaluate the rigor of triangulating across data voices; and introducing the term evidentiary heterogeneity to emphasize the value of diversifying data sources across cases.