A series of 237 patients with DSM-III-diagnosed panic disorder, or agoraphobia with panic attacks, received alprazolam as part of the placebo-controlled Cross-National Collaborative Panic Study. After a 1-week drug-free period, alprazolam dosage was titrated upward with the objective of reaching 6.0 mg/d in all patients. At week 3 of treatment, alprazolam plasma levels were significantly correlated with daily dosage (regression slope: 11.7 ng/mL per milligram per day) but with considerable individual variation. Among patients with spontaneous panic attacks, 70% of those with plasma alprazolam levels greater than 20 ng/mL achieved complete remission vs 31% of those with levels less than 20 ng/mL. Situational panic attack remission increased in frequency with increasing plasma levels, but the relationship was not significant. Patient- and physician-rated global improvement and Hamilton Anxiety and Depression Scale score reductions were maximal at 20 to 39 ng/mL, with no further benefit at higher levels. Central nervous system-depressant side effects increased in frequency with higher plasma levels. Between weeks 3 and 8 of treatment, physicians were permitted to adjust dosage (maximum: 10 mg/d) to optimize response. At week 8, the dose-concentration relationship was essentially identical (regression slope: 10.8 ng/mL per milligram per day), but plasma levels were no longer related to efficacy or side effects. Thus, monitoring of plasma alprazolam concentrations may have a clinically useful role during short-term treatment of panic disorder.