Violent extremism research is still lacking a sound empirical basis for the validation of assessment instruments. Yet there is a growing need for these instruments to assess the dangerousness of individuals, but also the success of interventions. By analysing prisoner files of one female and 39 male inmates (average age 28.83 years, SD = 7.58) with administratively assigned Islamism-related security labels in Bavarian prisons, we tried to clarify two questions: Firstly, is it possible to collect relevant data from prisoner files drawing on risk assessment procedures? Secondly, how do inmates associated with the Salafist scene (security label "Salafist scene") differ from those who are apparently involved with terror networks (security label "terror"), and do these differences predict the risk they pose? Our results suggest that files are a valuable, though not perfect data source for individual assessment and research. The two groups defined by the labels differ significantly in their biographies, mental health, and behaviour. Conclusions pertaining to biographical background factors, risk assessment, and management are discussed.