<p>Studies on temporal cognition show our sense of subjective time is biased by both attention and emotion: Duration estimates increase when attention is engaged by timing and by arousing stimuli. However, arousing stimuli themselves tend bias attention, complicating the degree to which experiments can dissociate cognitive and emotional factors. The present study therefore employed an affective imagery task to manipulate arousal and valence while participants engaged in a verbal estimation task, timing a neutral visual stimulus. Interestingly, while valence had a modest effect on interval estimates – negative affect increasing subjective time – arousal did not affect duration estimation. However, the more metacognitive indication of how swift participants experienced the passage of time showed pronounced effects of both arousal and valence, with arousal additionally increasing the difference between positive and negative affective states. While the effect of valence on time estimates replicated the broader literature, we argue the apparent absence of arousal-induced temporal inflation resulted from attentional requirements being equal across emotional scenarios. Judgements of the passage of time, conversely, are more likely to result from demand characteristics altering recall or informing decision-making. Such judgements may account for previously reported effects of arousal affecting perception of longer time intervals, rather than a speeding up of a hypothetical central pacemaker.</p>