There have been six randomized controlled trials of psychological therapy with generalized anxiety disorder (GAD) using DSM-III-R and DSM-IV. All have used the Trait version of the Spielberger State-Trait Anxiety Inventory (STAI-T) as one of several outcome measures. Each study, however, employed different methods of calculating the clinical significance of outcomes making it difficult to reach a balanced appraisal of the efficacy of psychological treatment. Raw data on STAI-T scores at pre-, post- and follow-up were obtained for each of the six studies (total N = 404). Jacobson methodology for defining clinically significant change (criterion c, reliable change index = 8, cut-off point = 46) was used to allocate each patient to one of four outcomes: worse, unchanged, improved and recovered. The proportion of patients in each category was calculated for treatment conditions in each study and also for aggregate data across types of treatment. A recovery rate of 40% was found for the sample as a whole with 12 of the 20 treatment conditions obtaining very modest recovery rates of 30% or less. Two treatment approaches--individual cognitive behavioural therapy and applied relaxation--do relatively well with overall recovery rates at 6-month follow-up of 50-60%. Jacobson methodology, in distinguishing between improvement and recovery on a standardized measure of general vulnerability to anxiety, provides a stringent but clinically more meaningful evaluation of the efficacy of psychological therapies with GAD than has been available hitherto. Systematic focus on either excessive worry or physiological arousal gives worthwhile results.