Purpose: To assess by analysis of published controlled trials the efficacy of cognitive behavioral therapies (such as biofeedback, relaxation, meditation) for essential hypertension. Data identification: Randomized controlled trials published in the English language between 1970 and 1991 identified from the MEDLINE database and bibliographic references from these articles. Study selection: Limited to studies involving randomized assignment to a treatment group consisting of one or more cognitive behavioral interventions or a concurrent control group consisting of no therapy, a waiting list, regular monitoring, or placebo intervention. Results of data synthesis: Although we identified more than 800 published works, only 26 met entry criteria. We identified a number of methodologic short-comings, including small sample size, inconsistencies regarding baseline blood pressure determinations and types of control groups, and the possibility of confounding by multiple noncognitive cointerventions (diet, exercise) and expectancy (the placebo effect). In meta-analyses involving 1264 patients, differences in mean blood pressure reduction varied according to the duration of baseline blood pressure measurements and type of control groups studied. In 16 comparisons involving baseline periods of more than 1 day, with patients (n = 368) assigned to either a cognitive therapy or a placebo intervention (sham biofeedback, "pseudo-meditation"), systolic and diastolic blood pressures decreased by 2.8 mm Hg (95% CI, -0.8 to 6.4) and 1.3 mm Hg (CI, -1.3 to 3.8), respectively. These changes were neither statistically nor clinically significant. Conclusion: Cognitive interventions for essential hypertension are superior to no therapy but not superior to credible sham techniques or to self-monitoring alone. The literature on this subject is limited by a variety of methodologic inadequacies. No single cognitive behavioral technique appears to be more effective than any other.