Contemporary criminal phenomena, such as terrorism and its financing, are based on patterns only to a certain extent in terms of execution. In reality, perpetrators act unconventionally, changing the tactics of action but also the patterns of the terrorist cells themselves. Thus, it is increasingly difficult to conduct reconnaissance of this criminal activity, because knowledge about it is limited (hidden) and, in addition, there is a time delay in the cause-effect analysis. Thus, new solutions from the field of mathematics are reached for. One of them is the use of hidden Markov models, the properties of which allow building predictive models based on limited knowledge resources, characterization of elements and assessment of the probability of events. Their application is wide and concerns not only the recognition of physical acts of terrorism, but also their financing or money laundering.