Matching social support to the recipient's needs requires diagnostic sensitivity on the part of the provider. In particular, support needs to be responsive to the recipient's stress-related appraisals to be maximally effective. To assess the impact of bias in interpersonal stress assessment, medical students in 43 dyads reported on their own and each other's stress appraisals, social support, affect and performance during a 5-day preparation period culminating in a multiple choice examination. Less biased perceptions of loss appraisals by support providers within dyads were followed by support transactions associated with lower negative affect and better exam performance among recipients. More biased perceptions of threat appraisals were followed by increases in the recipients' negative affect. Results therefore suggest that support is more effective when the provider understands the recipient's concerns.