Background: Educational attainment is associated with wellbeing and health, but patients with schizophrenia achieve lower levels of education than people without. Several effective interventions can ameliorate this situation. However, the magnitude of the education gap in schizophrenia and its change over time are unclear. We aimed to reconstruct the trajectories of educational attainment in patients with schizophrenia and, if reported, their healthy comparator controls. Methods: We did a systematic review and meta-analysis including all studies reporting on patients with schizophrenia (of mean age ≥18 years) and describing the number of years of education of the participants, with or without healthy controls. There were no other design constraints on studies. We excluded studies that included only patients with other schizophrenia spectrum disorders and studies that did not specify the number of years of education of the participants. 22 reviewers participated in retrieving data from a search in PubMed and PsycINFO (Jan 1, 1970, to Nov 24, 2020). We estimated the birth date of participants from their mean age and publication date, and meta-analysed these data using random-effects models, focusing on educational attainment, the education gap, and changes over time. The primary outcome was years of education. The protocol was registered on PROSPERO (CRD42020220546). Findings: From 32 593 initial references, we included 3321 studies reporting on 318 632 patients alongside 138 675 healthy controls (170 941 women and 275 821 men from studies describing sex or gender; data on ethnicity were not collected). Patients’ educational attainment increased over time, mirroring that of controls. However, patients with schizophrenia in high-income countries had 19 months less education than controls (–1·59 years, 95% CI –1·66 to –1·53; p<0·0001), which is equivalent to a Cohen's d of –0·56 (95% CI –0·58 to –0·54) and implies an odds ratio of 2·58 for not completing 12 years of education (ie, not completing secondary education) for patients compared with controls. This gap remained stable throughout the decades; the rate of change in number of total years of education in time was not significant (annual change: 0·0047 years, 95% CI –0·0005 to 0·0099; p=0·078). For patients in low-income and middle-income countries, the education gap was significantly smaller than in high-income countries (smaller by 0·72 years, 0·85 to 0·59; p<0·0001), yet there was evidence that this gap was widening over the years, approaching that of high-income countries (annual change: –0·024 years, –0·037 to –0·011; p=0·0002). Interpretation: Patients with schizophrenia have faced persistent inequality in educational attainment in the last century, despite advances in psychosocial and pharmacological treatment. Reducing this gap should become a priority to improve their functional outcomes. Funding: Ciencia y Tecnología para el Desarrollo (CYTED) to the Latin American Network for the Study of Early Psychosis (ANDES).