This study examined the influence of serial position of subjects’ affect ratings of stimuli. Ninety subjects saw a serial list of 12 nonsense words repeated two, four, or eight times. Subjects rated the words and then attempted to recall them. Results indicated that: (a) A curvilinear relationship exists between serial position and affect, with first and last items preferred to middle items. (b) Recall is most accurate for the first items in the list. (c) Partially and completely recalled words are preferred to words which were not recalled. (d) After controlling for recall, the curvilinear relationship between serial position and affect still remains.