Résumé : Although standardized role-playing sessions (SRPS) with simulators are increasingly used to assess health care professionals' (HCPs) communication skills (CS) and the effectiveness of training workshops (TWs), nothing has been done to date to define the optimal emotional content of SRPS. Three emotionally different SRPS contexts-weakly emotional (WE-), moderately emotional (ME-), and highly emotional (HE-SRPS)-were, therefore, tested in order to assess induced CS and sensitivity to TW-related changes. The study included 25 HCPs. Tape-recorded SRPS, scheduled before and after the TW, were retranscribed, and assessed according to the Cancer Research Campaign Workshop Evaluation Manual (CRCWEM), which provides a rating of form, function and structure for each utterance. Results show that induced CS are different in WE-, ME-, and HE-SRPS, regarding form (HE-SRPS induced more 'directing', 'leading' or 'multiple' questions; WE: 20.7%; ME: 19.7%; HE: 33.7% (p<0.001)); function (HE-SRPS induced more 'inappropriate' information; WE: 6. 5%; ME: 8.2%; HE: 15.6% (p<0.001)); and blocking (HE-SRPS induced more 'blocking' utterances; WE: 7.2%; ME: 13.8%; HE: 30.2% (p<0. 0001)). Finally, CS changes induced by TWs are the highest in HE-SRPS (14.8% increase of 'open' questions for the HE- versus 1.0% for the WE-SRPS; 11.6% decrease of 'inappropriate' information for the HE- versus 3.3% for the WE-SRPS; and 17.5% decrease of 'blocking' for the HE- versus 2.6% for the WE-SRPS). In conclusion, SRPS, with a HE content, induce more inappropriate CS. Moreover, they are more sensitive to TW effects. SRPS with a HE content should, thus, be recommended for the assessment of TW effectiveness.