Background Lithium has long been recognised as an effective treatment for bipolar disorder. Its relative efficacy has been measured with a diverse range of clinical outcomes, resulting in differences in efficacy reporting that have not been systematically reviewed. Aims We aimed to identify and compare the various measures of lithium efficacy employed in interventional studies for people with bipolar disorder. Method Database (PubMed, Web of Science) and hand searches were performed to identify studies that assessed a clinical response in patients with bipolar disorder who received lithium, up to the end of 2021. We included primary human interventional studies without excluding specific study designs, bipolar disorder subtypes, duration or dosage of lithium treatment. Continuous outcome effects were meta-analysed; binary outcomes were synthesised visually and narratively. The Cochrane risk-of-bias tool was used to assess study-level risk of bias. Results Seventy-one studies were included (N = 30 542). Approximately two-thirds of participants attained a clinically significant improvement in manic or depressive symptoms, and over 50% achieved remission. About a third required hospital admission (study length 2–12 years) and around 50% needed further treatment to stay well or had recurrence of symptoms; the latter two outcomes tended to be assessed over long-term maintenance periods. Conclusions An abundance of measurements have been used to assess lithium's clinical effects, across several study designs. Despite the resultant high heterogeneity, an overall picture of lithium's effects emerges that supports previous literature; between half and two-thirds of patients respond well to lithium across varying outcome measures, baseline mood states, study durations and bipolar disorder subtypes.