AbstractWe evaluated a newly developed stress task, the Social Competence Interview, and three nonsocial tasks (video game, mirror drawing, mental arithmetic) for ability to predict ambulatory blood pressure in 237 black and white adolescents. Blood pressure was measured in laboratory, classroom, and transition (between‐class) settings. A resting laboratory baseline explained 10–49% of the variance in ambulatory blood pressure levels; the ability of the stress tasks to explain additional variance was assessed in multiple regression analyses. Only the blood pressure response to the interview enhanced prediction of classroom and transition systolic and diastolic pressures in the total sample and in blacks, whites, females, and males – even when the interview data were entered into a hierarchical regression model after those for the other three tasks were entered. Mirror drawing improved prediction of transition systolic blood pressure in the total sample, and mental arithmetic plus the interview improved prediction of classroom diastolic pressure in black males; however, video game failed to enter any predictive equation. Racial subgroup analyses disclosed that the interview data predicted systolic pressure in whites but predicted diastolic pressure in blacks, indicating biological differences in blood pressure regulation. An interview that elicits characteristic thoughts and social behaviors appears to represent a promising approach to examining environmental influences on blood pre