From three meta‐analyses on treatment outcome in panic disorder with agoraphobia, method, results and conclusions were reviewed. The meta‐analyses differed in methods used and conclusions derived. The conclusions conflicted on the relative efficacy of treatment with antidepressants, high‐potency benzodiazepines, psychological panic management, exposure in vivo and combination treatments. These conflicting findings can be caused by confounding variables, present in ‘between‐study’ comparisons. In ‘within‐study’ comparison, so‐called ‘same experiment studies’, these confounding variables are controlled for. Therefore, the literature was reviewed for same experiment studies. The differential efficacy of treatments compared within 25 identified same experiment studies was evaluated by calculating the effect size d between the treatments. Antidepressants and high‐potency benzodiazepines were equally effective. The comparison between these drugs and psychological panic management remained inconclusive. Exposure in vivo was superior to psychopharmacological and psychological panic management. The efficacy of exposure in vivo was not enlarged by the addition of psychological panic management. There was some evidence that the addition of psychopharmacological drugs to exposure in vivo enlarged the efficacy of the latter. This observation must be investigated more systematically.