The therapeutic alliance, or client-provider relationship, has been associated with better treatment engagement and outcomes for persons with schizophrenia-spectrum disorders (SSDs) and early psychosis in some studies, but not others. We conducted a meta-analysis of the research on alliance in SSDs and early psychosis across a range of interventions and outcomes. Parallel literature searches were conducted in PubMed and PsycINFO databases for articles between inception and 6/11/2020. English-language studies were included if they evaluated the relationship between alliance and a prospective outcome (treatment engagement, medication adherence, functioning, or total, positive, negative, or depressive symptoms) in an individual clinical treatment for SSDs/early psychosis and contained analyzable data. Correlations and partial correlations were meta-analyzed with random effects models to calculate mean across-study correlations and to carry out subsequent homogeneity and moderator variable analyses. Fourteen studies consisting of 2968 participants that assessed six outcomes across six psychosocial treatments were included. Results indicated that better client-rated (r = 0.20) and other-rated (i.e., provider- or observer-rated; r = 0.25) alliance were associated with better treatment engagement. Treatment type and sample race/ethnicity, but not age, gender, or timing of alliance rating moderated the association between other-rated alliance and engagement. Further, better other-rated alliance was related to improvements in positive (r = -0.14) and negative (r = -0.22) symptoms. A strong therapeutic alliance is important for both engaging clients with SSDs and early psychosis in treatment and facilitating improvements in positive and negative symptoms. Delivery and monitoring of treatments for this population should include assessment of the therapeutic alliance from multiple perspectives.