Two hundred and two moderately chronic psychiatric outpatients, all suffering from anxiety of at least moderate severity and all diagnosable as cases of Generalized Anxiety Disorder, participated in a single-blind 6-week trial of diazepam (15-40 mg/day). The trial was preceded by a 1 week placebo washout, and provided for evaluation visits after 1, 2, 4 and 6 weeks of diazepam treatment. Patients were divided into High, Medium and Low Initial Improvers using 1 week change in Hamilton Anxiety Scale total score to assign patients to three subgroups of equal size. These groups did not differ significantly on those demographic factors and attributes of illness history which were documented, nor on assessments of symptom and illness severity, and mode of intake. Examination of a number of patient and physician assessments of illness severity revealed that the High group had the greatest 6-week improvement, the Low group the least. During the first week, the High group attained 86%, the Medium group, 65%, and the Low group, 29% of its full 6-week drug response. Diazepam dose levels were lowest for the High group and highest for the Low group. Placebo response was least for the High group and greatest for the Low group. An attempt to find distinctive attributes of the three initial improvement groups was unsuccessful.