Summary The effects of the evaluative connotation of the context on free recall and on evaluative judgments were assessed experimentally. The materials were four lists of common nouns which had been previously rated for pleasantness. The lists were Neutral, Heterogeneous, Good, and Bad, and the sets of words in these lists were, respectively: Neutral, Moderately Good, and Moderately Bad; Good and Bad; Good, Moderately Good, and Neutral; Bad, Moderately Bad, and Neutral. The lists were so constructed that various sets of words appeared in two or more lists. Total recall was greatest from an Heterogeneous List, followed by Good and Neutral Lists, with recall from a Bad List significantly lower than from the other lists. On the whole, Good words were recalled best, followed by Bad, and finally, Neutral words. However, in a Neutral List overall superiority in recall of Moderately Good over Moderately Bad words was significant only for intentional learners. Moderately Bad words were recalled better in a Neutral List than in a Bad List, and there was a trend for Neutral words to be recalled better in Good and Neutral Lists than in a Bad List. Neutral words were also better recalled under incidental than under intentional set. The evaluative judgments showed clear contrast effects, i.e., Good words were rated as more pleasant in an Heterogeneous List than in a Good List; Bad words were rated less pleasant in the Heterogeneous List than in the Bad List. Similarly, Moderately Bad words were rated significantly less pleasant in the Neutral List than in the Bad List. Although the ratings of the Neutral words did not vary significantly among lists, the within-subject variability of the ratings was significantly greater in the Neutral List than in the Good or Bad Lists.