AIMS: Although the relationship between body mass index and all-cause mortality is U-shaped, underweight has received comparatively less attention than obesity. There is only limited evidence to date regarding underweight among patients with schizophrenia. This is the first meta-analysis to address the prevalence of underweight in these patients. METHODS: We conducted database searches (PubMed, PsycINFO) to identify studies examining underweight in patients with schizophrenia. In total, 17 studies (18 groups) with 45,474 patients were included; data were extracted independently by two authors. A meta-analysis was performed to calculate the pooled prevalence of underweight in patients. RESULTS: The pooled prevalence of underweight was 6.2% (95% CI=4.5-8.6) for the 18 groups, which included 45,474 patients with schizophrenia. The heterogeneity was I2=98.9% (95% Cl=98.7-99.1%). Four studies with 4 groups, consisting of 30,014 individuals, focused on Japanese inpatients with schizophrenia. The pooled prevalence of underweight among inpatients in these 4 groups was 17.6% (95% CI=15.5-20.0). Fourteen studies were conducted with non-Japanese inpatients and included 14 groups of 15,460 patients with schizophrenia. The pooled prevalence of underweight in non-Japanese inpatients was 4.6% (95% CI=3.9-5.4). The proportion of underweight in the 18 groups significantly varied between Japanese inpatients and other patients. CONCLUSIONS: The results indicated that Japanese inpatients with schizophrenia have a high proportion of underweight. Future research should focus on evaluating interventions that target underweight.