We report a meta- and primary data-analysis investigating the efficacy of cognitive therapy (CT) for pathological worry in adults with GAD that includes an analysis of primary data not reported in previous meta-analyses. Eligible studies included those whose participants met the criteria for a diagnosis of GAD and those that used the PSWQ as an outcome measure. All eligible studies used a randomized controlled design. Analyses included a random-effects meta-analysis of between-study effect sizes and hierarchical linear models of both within study change over time and primary recovery data. The results show that CT was effective in reducing pathological worry when compared with non-therapy controls (d=1.81), and gains were largely maintained at follow-up. The magnitude of effects reported was larger than previously found, suggesting an increased efficacy of newer forms of CT. However, we found weaker evidence to suggest that CT for pathological worry was superior to non-CT treatment controls (d=0.63). Analysis of primary recovery data revealed that 57% of participants were classed as recovered at 12 months following CT, and CT had significantly better recovery rates than all other comparison treatments at post-treatment and 12-month follow-up. These findings support the increasing efficacy of CT as a treatment for GAD. However, CT interventions still need further refinement to help a greater proportion of sufferers achieve recovery. Copyright © 2012 Elsevier Ltd. All rights reserved.