The present study. using a between-participants factorial ANOVA design, investigated the extent to which self-generated interactive images (standard keyword) were more effective than researcher-supplied images (keyword + interaction) in the long-term retention of EFL vocabulary by college students. It also assessed the extent to which these two versions of the keyword method were more effective than the rote method. The results indicated that the effectiveness of self--generated images was statistically comparable to that of researcherdeveloped ones. Additionalll, the results indicated that only, the standard versión of the keyword method produced significantly superior long-term recall to the rote method.