This study determines patient characteristics that predict early hospital readmission in schizophrenia and evaluates the extent to which inpatient staff accurately predict these readmissions. Adult inpatients with schizophrenia or schizoaffective disorder (N = 262) were evaluated at hospital discharge and 3 months later to assess hospital readmission. At hospital discharge, inpatient staff were asked to identify which patients were likely to be readmitted during this period. Comparisons were made between patients who were or were not readmitted and between readmitted patients who were or were not identified by staff as likely to be readmitted; 24.4% of the sample were readmitted within 3 months of hospital discharge. Early readmission was associated with four or more previous hospitalizations (85.7% vs. 57.7%, p = .004), comorbid substance use disorder (60.3% vs. 35.5%, p = .0006), major depression (40.6% vs. 26.8%, p = .04), absence of a family meeting with inpatient staff (58.2% vs. 41.8%, p = .02), and prescription of a conventional rather than an atypical antipsychotic medication (93.7% vs. 83.8%, p = .045). Twelve of the 63 readmitted patients were correctly predicted by staff to re-hospitalize. Staff tended to overestimate the risk of rehospitalization in patients with a poor therapeutic alliance, low global function, or initial involuntary admission and to underestimate the risk in patients with alcohol use disorders or four or more previous psychiatric hospitalizations. Early rehospitalization is common in schizophrenia and difficult to predict. Greater emphasis on comorbid alcohol use disorders and a history of multiple previous admissions may help clinicians identify patients at greatest risk for early rehospitalization.