Affective priming studies have shown that participants are faster to pronounce affectively polarized target words that are preceded by affectively congruent prime words than affectively polarized target words that are preceded by affectively incongruent prime words. We examined whether affective priming of naming responses depends on the valence proportion (i.e., the proportion of stimuli that are affectively polarized). In one group of participants, experimental trials were embedded in a context of filler trials that consisted of affectively polarized stimulus materials (i.e., high valence proportion condition). In a second group, the same set of experimental trials was embedded in a context of filler trials consisting of neutral stimuli (i.e., low valence proportion condition). Results showed that affective priming of naming responses was significantly stronger in the high valence proportion condition than in the low valence proportion condition. We conclude that (a) subtle aspects of the procedure can influence affective priming of naming responses, (b) finding affective priming of naming responses does not allow for the conclusion that affective stimulus processing is unconditional, and (c) affective stimulus processing depends on selective attention for affective stimulus information.