Participants scoring high and low on a measure of repressive coping style (Mendolia, 2002) first learned a series of related word pairs (cue-target). Half of the cues were homographs. In the subsequent think/no-think phase (Anderson & Green, 2001), they responded with targets on some trials and suppressed thoughts of targets on others. Suppressed targets were always emotionally negative, as were targets associated with baseline cues reserved for the final test. Some participants were provided with emotionally benign or positive substitutes to help them suppress, and these substitutes were related to different meanings of the homographic cues, compared to those established by the targets. On the final test, all cues were presented for target recall. Only the repressors significantly benefited from the provision of positive substitutes to aid forgetting of the negative targets, regardless of the nature of the cues.