A better understanding of how social factors relate to the psychological processes in thought disorder (TD) is necessary for the development of effective psychological interventions. Sixty-eight participants diagnosed with psychosis (18–65; 47.1% female) were recruited and evaluated on social cognition (Hinting Task, HT; and reading the mind in the eyes test, RMET), social isolation (size of social network, frequency and quality of contact), psychotic symptoms (Positive and Negative Syndrome Scale, PANSS) and TD (Thought, Language and Communication Disorders Scale, TLC). A mediation model was tested with isolation as the predictor, TD as the outcome, and performance on HT and RMET as the mediators. The final model, with adjustment for comorbid symptoms (i.e. delusions, suspiciousness, hallucinations, and negative symptoms), supported full mediation and explained a significant amount of the observed variance (60%). Performance on the HT was a significant mediator of the relationship between social isolation and TD. From the covariates, delusions contributed independently and significantly to TD. The implications of the findings for psychological practice, and TD-specific interventions, are discussed as well as the limitations of the study. Further avenues for symptom-specific research are discussed, in particular with reference to more complex psychosocial models. © 2018