The present study investigated the effect of gender-related words on source monitoring in identifying a speaker's gender. During an acquisition phase, participants were presented with a set of target words spoken by a man and a woman. The word list included three categories of words; male-related words (e.g. rugby), female-related words (e.g. skirt), and gender-neutral words (e.g. eraser). The words were presented in three conditions: a congruent condition (male-related words spoken by a male voice, and femalerelated words spoken by a female voice), an incongruent condition (female-related words spoken by a male voice and male-related words spoken by a female voice), and a control condition (gender-neutral words spoken by a male and a female voice). The word stimuli for half of the participants were presented in an audio-only condition (through an audio player), and an audio-visual condition (through a video of the speaker saying the words) for the other half. During the testing phase, participants completed a source-monitoring test in which they were asked to identify the source of each word among three alternatives (man, woman, neither). The results indicated that source monitoring was more accurate in the incongruent condition compared to the congruent and control conditions. There were no differences between the audio-only and audio-visual conditions. These results are discussed from the perspective of the level of source attribution.