The aim of this study was to compare a traditional stress setting, consisting of two mental arithmetic tasks and two Stroop test modifications, and a stress setting of varying task demand and decision latitude according to Karasek's job strain model, with respect to their feasibility to elicit differences in cardiovascular reactivity and recovery in 20 normotensives, 20 borderline hypertensives, and 20 non-medicated hypertensives, carefully selected by means of World Health Organization criteria. In addition, the relationship between laboratory and everyday blood pressure was investigated. All subjects were tested under both stress settings in counterbalanced order. Blood pressure was recorded both intermittently from the brachial artery (Riva-Rocci) and continuously from the finger (Finapres). Heart rate and electrodermal activity were continuously measured as well. Furthermore, daily life blood pressure recorded by means of 24 h ambulatory monitoring during a normal working day served as criterion for the re-classification of the blood pressure groups by means of discriminant analysis using physiological recordings from baseline, test phases and rest phases. The groups did not show significant differences in their reactivity to the various mental stressors including the Karasek-model oriented ones but marked differences in their behaviour occurred during the 10 min of recovery following each stress setting. Both systolic and diastolic blood pressure in hypertensives failed to recover during this period. The results also showed the superiority of the Finapres method with respect to reflecting the dynamics of physiological recovery processes. None of the stress settings showed an advantage in predicting blood pressure in daily life. In general, the results question the validity of mental laboratory stressors for the prediction of cardiovascular changes in daily life but point to a possible role of recovery processes after stress in the development of essential hypertension.