INTRODUCTION: The objective was to evaluate efficacy/safety of complementary and alternative medicine (CAM) methods for generalized anxiety disorder (GAD) based on randomized controlled trials in adults. METHODS: Data sources. Six electronic databases ("generalized anxiety (disorder)" and "randomized trial") and reference lists of identified publications were searched to March 2017. STUDY SELECTION: Eligibility: full-text publications (English, German language); CAM versus conventional treatment, placebo/sham or no treatment; GAD diagnosed according to standard criteria; and a validated scale for disease severity. Of the 6693 screened records, 32 were included (18 on biologically-based therapies, exclusively herbal preparations; eight on manipulative and body-based therapies; and three on alternative medical systems and three on mind-body therapies). DATA EXTRACTION: Cochrane Collaboration methodology was used for quality assessment and data extraction. RESULTS: Direct comparisons of Kava Kava (Piper methysticum) extracts to placebo (4 quality trials, n = 233) were highly heterogeneous. Network meta-regression reduced heterogeneity and suggested a modest Kava effect [end-of-treatment Hamilton Anxiety scale score difference adjusted for baseline scores and trial duration: - 3.24 (95% CI - 6.65, 0.17; P = 0.059), Kava Kava 4 arms, n = 139; placebo 5 arms, n = 359]. Lavender (Lavandula angustifolia) extract (1 quality trial, 10 weeks, n = 523) and a combination of extracts of C. oxycantha, E. californica and magnesium (1 quality trial, 12 weeks, n = 264) were superior to placebo and balneotherapy was superior to paroxetine (1 quality trial, 8 weeks, n = 237) indicating efficacy. All other trials were small and/or of modest/low quality and/or lacked assay sensitivity. Safety reporting was poor. CONCLUSION: Evidence about efficacy/safety of most CAM methods in GAD is limited. Apparent efficacy of certain herbal preparations and body-based therapies requires further confirmation.