The purpose of this study was to determine whether affect influences mediate association. A second purpose of this experiment was to test whether there could be found an interaction between affect and meaningfulness in the verbal mediation scores. The subjects were all of the students registered for an Educational Psychology class at Utah State University, Spring Quarter, 1969. These students were randomly assigned to one of two groups, designated Phase I or Phase II. Phase I was designed to study the influence of affect upon mediation and the subjects in this group learned two lists of seven paired associates. Phase II was designed to examine the possible interaction of affect and meaningfulness in mediation and the subjects in this group learned two lists of eight paired associates. Phase III was added to the study to determine if there would be a correlation between mediation and association ability of all of the subjects. Affect level was determined by the magnitude of the Galvanic Skin Response readings on Stoelting Psychogalvanoscope in reaction to the mediating words of the B list. Meaningfulness level of the non-mediators was defined as the association value of Consonant-Vowel-Consonant trigrams used in the A-C lists. Mediation was defined as the number of correctly paired A-C trigrams in the multiple-choice mediation test. To test whether affect influences mediation, a comparison was made between mediation scores produced by high affect mediators and mediation scores produced by low affect mediators. The test of the interaction was made by a factorial design with two levels (high, low) of affect and four combinations of levels (high-high, high-low, low-high, and low-low) of meaningfulness. The procedure first assessed the affect level of the mediators. Then either Phase I, which tested Hypothesis 1, or Phase II, which tested Hypothesis 2, was administered to each subject. Each phase followed the chaining model (A-B, B-C, A-C) of mediation. There was no learning of the A-C list, but mediation was tested by pairing the A-C items in a multiple-choice test. Also, a test of association ability was made after presenting twelve paired associates using a similar multiple-choice test to that used to test mediation. Statistical analyses were applied to these test scores to determine the empirical support of the following hypotheses: Hypothesis 1 proposed that there would be a significant difference between the amount of recall scores mediated by high and low affect words when the meaningfulness of the non-mediators is held constant at a medium level. This hypothesis was supported by the data obtained. Hypothesis 2 predicted that there would be an interaction between levels of affect and combinations of levels of meaningfulness. This hypothesis was strongly supported by the data of this study. An additional finding was that a low, but significant correlation was obtained between mediation scores and association scores. The findings of this study showed that affect level of the mediator affects the amount of mediation produced in a chaining paradigm. There appears to be strong evidence for an affect and meaningfulness interaction in mediation data. Within this interaction, there was an indication that affect is prepotent over meaningfulness. Also, analysis of this interaction shows that the meaningfulness of the stimulus term rather than the response term seems to be critical in producing superior mediation. Finally, a low correlation seems to exist between simple or paired association and mediate association, because simple (paired) association and mediate association do not seem to be identical processes.