This paper discusses the use of computer support to analyze IS propaganda. We focus on thematic analysis of the textual content of various types of online propaganda, and investigate two different ways to identify themes in the data. One approach is theory-driven and builds on previous analysis of IS propaganda, and the other approach is data-driven and automatically identifies prevalent themes in the data. We exemplify the various approaches using different types of IS propaganda and we also provide examples of how the computer-assisted analysis can be used to analyze variation in the data over time. Both approaches indicate an increase in more violent themes in more recent IS propaganda.