Background Although the efficacy of Internet‐ and mobile‐based interventions (IMIs) for anxiety is established, little is known about the intervention components responsible for therapeutic change. We conducted the first comprehensive meta‐analytic review of intervention components of IMIs for adult anxiety disorders. Methods Randomized controlled trials (RCTs) comparing IMIs for anxiety disorders to active online control groups, or IMIs to dismantled variations of the same intervention (± specific components) were identified by a systematic literature search in six databases. Outcomes were validated observer‐rated or self‐report measures for anxiety symptom severity and treatment adherence (number of completed modules and completer rate). This meta‐analytic review is registered with PROSPERO (CRD42017068268). Results We extracted the data of 34 RCTs (with 3,724 participants) and rated the risk of bias independently by two reviewers. Random‐effects meta‐analyses were performed on 19 comparisons of intervention components (i.a., different psychotherapeutic orientations, disorder‐specific vs. transdiagnostic approaches, guidance factors). IMIs had a large effect when compared to active online controls on symptom severity (standardized mean difference [SMD] of −1.67 [95% CI: −2.93, −0.42]; P = 0.009). Thereby, guided IMIs were superior to unguided interventions on symptom severity (SMD of −0.39 [95% CI: −0.59, −0.18]; P = 0.0002) and adherence (SMD of 0.38 [95% CI: 0.10, 0.66]; P = 0.007). Conclusions Overall, the results of this meta‐analysis lend further support to the efficacy of IMIs for anxiety, pointing to their potential to augment service supplies. Still, future research is needed to determine which ingredients are essential, as this meta‐analytic review found no evidence for incremental effects of several single intervention components apart from guidance.