The research aimed to develop and test a new dynamic approach to preventive risk assessment of violent extremists. The well-known New York Police Department four-phase model was used as a starting point for the conceptualization of the radicalization process, and time-stamped biographical data collected from court documents and other public sources on American homegrown Salafi-jihadist terrorism offenders were used to test the model. Behavioral sequence patterns that reliably anticipate terrorist-related criminality were identified and the typical timelines for the pathways to criminal actions estimated for different demographic subgroups in the study sample. Finally, a probabilistic simulation model was used to assess the feasibility of the model to identify common high-frequency and high-risk sequential behavioral segment pairs in the offenders' pathways to terrorist criminality.