Social media platforms, such as Instagram, are regularly misused for spreading covert (Islamic) extremist propaganda. Affect and emotion are central tools used in extremist propaganda, but there is little research into the combined employment of different social media elements, such as hashtags, visuals, and texts, in the context of propaganda. This study contributes to closing this gap. Using the German group Generation Islam as a case study, we examined the group’s Instagram activity ( N = 1,187 posts) over the course of 2 years. To reflect the platform users’ logic, we (a) examined affect in hashtag networks in which users can come across propagandistic content, (b) employed deep learning to examine the emotional valence transmitted in the visuals, and (c) used automated linguistic analysis to describe collective action cues contained within the texts. The results are novel, as they provide nuanced insights into extremist propaganda’s employment of affect and emotions across Instagram’s affordances.