To model the dynamics of cognitive processes, often the dynamical systems theory (DST) is advocated. However, for higher cognitive processes such as reasoning and certain forms of natural language processing the techniques adopted within DST are not very adequate. This paper shows how an analysis of the dynamics of reasoning processes can be made using techniques different from those of DST. The approach makes use of temporal traces consisting of sequences of reasoning states over time to describe reasoning processes. It is shown for the example reasoning pattern 'reasoning by assumption', how relevant dynamic properties can be identified and expressed using a temporal trace language. Example traces have been acquired in two ways. Firstly, empirical traces have been generated based on think-aloud protocols of a subject solving a number of cases of a reasoning puzzle. Secondly, a simulation model has been developed and executed for the same cases of the reasoning puzzle. For all these traces, the dynamic properties can and have been analysed automatically, using a software environment that has been developed. Thus the feasibility of the approach was shown. © Elsevier Science B.V. All rights reserved.