It is a widespread opinion that after treatment with psychotherapy, patients with anxiety disorders maintain their gains beyond the active treatment period, whereas patients treated with medication soon experience a relapse after treatment termination.

We aimed to provide evidence on whether enduring effects of psychotherapy differ from control groups.

We searched 93 randomised controlled studies with 152 study arms of psychological treatment (cognitive–behavioural therapy or other psychotherapies) for panic disorder, generalised anxiety disorder and social anxiety disorder that included follow-up assessments. In a meta-analysis, pre-post effect sizes for end-point and all follow-up periods were calculated and compared with control groups (medication: n = 16 study arms; pill and psychological placebo groups: n = 17 study arms).

Gains with psychotherapy were maintained for up to 24 months. For cognitive–behavioural therapy, we observed a significant improvement over time. However, patients in the medication group remained stable during the treatment-free period, with no significant difference when compared with psychotherapy. Patients in the placebo group did not deteriorate during follow-up, but showed significantly worse outcomes than patients in cognitive–behavioural therapy.

Not only psychotherapy, but also medications and, to a lesser extent, placebo conditions have enduring effects. Long-lasting treatment effects observed in the follow-up period may be superimposed by effects of spontaneous remission or regression to the mean.

In the past 12 months and in the near future, Dr Bandelow has been/will be on the speakers/advisory board for Hexal, Mundipharma, Lilly, Lundbeck, Pfizer and Servier. Dr Wedekind was on the speakers' board of AstraZeneca, Essex Pharma, Lundbeck and Servier. All other authors have nothing to declare.