Cardiovascular effects of stress in humans are often assessed by application of physical or emotional stimuli in a laboratory environment. Although this method provides important information, these procedures have several limitations. First, blood pressure and heart rate responses to laboratory stressors are characterized by a limited within-subject reproducibility. Second, there is poor correlation between blood pressure and heart rate responses to different stressors, which implies that individual reaction to stress may be estimated differently according to the test used. Finally, these responses bear only a limited relation to 24-hour or daytime blood pressure variability, that is, they reflect to only a limited extent the tendency of blood pressure to vary during daily activities. If assessed by techniques that allow blood pressure to be continuously recorded for 24 hours in ambulatory subjects, blood pressure variability represents a possible approach to observation of cardiovascular reactivity away from an artificial laboratory environment. However, whether blood pressure variability should be expressed as a percentage or in absolute values is controversial. Furthermore, although naturally occurring stress may markedly increase blood pressure, 24-hour blood pressure variations also depend on factors that are not related to emotional stimuli. Thus, the study of cardiovascular responses to stress in humans encounters several problems, regardless of the method used.