Background: A common approach to assessing students' integrated understanding of multiple documents is to analyse their post-reading written reports. This study investigated to what extent writing self-efficacy directly and indirectly (via integrative processing) contributed to multiple-document comprehension as assessed with an integrative writing task. Methods: A sample of Norwegian university students (n = 67) read four documents on a controversial socio-scientific issue and afterwards wrote reports on the issue without access to the documents. Multiple-document comprehension was assessed in terms of how well the reports reflected an elaborated and integrated understanding of the four documents' content. A mediation analysis was conducted with students' working memory as a covariate, their confidence in their ability to write a text that integrated content from multiple source documents as a predictor, self-reports of their integrative processing during reading as a mediator and multiple-document comprehension as an outcome variable. Results: There was an indirect relationship between multiple-document-based writing self-efficacy and multiple-document comprehension via integrative processing. However, no direct relationship between writing self-efficacy and multiple-document comprehension was found. The covariate of working memory uniquely adjusted students' multiple-document comprehension. Conclusions: In the context of written assessment of multiple-document comprehension, multiple-document-based writing self-efficacy and multiple-document comprehension were indirectly related via integrative processing during reading. The results indicate that not only reading-related but also writing-related individual differences may come into play when multiple-document comprehension is assessed with an integrative writing task.