As the Affect Misattribution Procedure (AMP) is often used to measure interindividual differences in attitudes, a crucial question is how well the AMP reflects differences in the strength of attitudes. We approached this question by examining to what extent the AMP reflects gradual differences in prime valence. We also explored whether affective processes play a role in this regard. We used a modified AMP with a scalar response format in all experiments. Gradual differences between primes were generated in two ways: by selection of pictures according to normative ratings and by emotion regulation. In Experiment 1 (n = 55), we varied prime valence by selecting neutral, moderately and extremely positive pictures based on normative ratings from the IAPS. We additionally varied the arousal of the primes. In Experiments 2 (n = 81) and 3 (n = 142), participants first downregulated the intensity of their affective responses to positive or negative IAPS pictures by different emotion regulation strategies (distancing and reinterpretation). In a subsequent AMP, these pictures were used as primes. We found that the AMP reflected gradual differences of the prime valence when manipulated via the selection of stimuli according to normative ratings, whereas we found no reduction of AMP priming effects when the affective intensity of the primes was reduced by emotion regulation. Arousal did not play a role, presumably because the targets were to be evaluated in terms of valence. We assume that fast evaluations of the prime valence, retrieved from memory, determine AMP responses.