A single-blinded, randomized trial was conducted to determine whether a mailed postcard improved follow-up in uncontrolled hypertensives. One hundred and seven patients with a systolic blood pressure (BP) of 180 or more or a diastolic blood pressure of 110 or more at an inner-city, hospital-based walk-in clinic were enrolled; mean age was 56 years, 95 percent were African American, 73 percent were female, and mean BP was 193/106. Patients were required to be aware of their diagnosis and to have been informed of their need for medication at least a month before the trial. Of those who received postcard reminders, 45 percent followed up within 10 days, compared with 47 percent of controls (p = 0.93). At 30 days, 64 percent of the intervention group followed up, compared with 55 percent of controls (p = 0.36). In an adjusted logistic regression model, there was no difference in follow-up. Correlates of appointment noncompliance at one month included alcoholism and lack of insurance in an adjusted logistic regression model. Follow-up in severe hypertensives was poor, and a mailed postcard reminder had no effect in a walk-in setting.