<p>The current debate about mental imagery revolves around the puzzling fact that a sizable minority of people can do the sorts of task most of us perform with the help of mental imagery, but profess being unaware of any imagery in their own cases as they perform these tasks. The question is what to say about such ‘non-imagers’, and what to say about the utility of conscious imagery as a result. For example, if non-imagers lack imagery altogether, or are utilising unconscious imagery, the threat emerges that conscious imagery may simply be epiphenomenal as regards task performance: a ‘cognitive luxury’. This paper contributes to the debate by advancing a new position, and arguing that it is preferable to notable alternatives defended by Eric Schwitzgebel (2011), Ian Phillips (2014), and Bence Nanay (2021). The novel suggestion is that just as conscious mental imagery features mental qualitative properties in virtue of which it contributes to task performance (‘mental qualities’), unconscious mental imagery also features relevant mental qualities, albeit in unconscious form. This view, it is argued, has the best prospects of safeguarding various plausible theses about mental imagery and its role in mental life. The alternative views, by contrast, all involve giving up either the causal relevance of conscious imagery, or our introspective access to it.</p>