The study investigates with inductive content analysis the verbal manifestations of schizophrenia patients, their utterances within the theory of mind modules of Metacognitive Training. Participants were recruited via psychiatrists’ verbal referrals. Four small groups were formed, two of which consented to the audio recording of their sessions. Only the sessions of these two groups of five and four patients respectively were investigated. As two participants dropped out, the verbalisations of seven participants were eventually analysed. All patients had schizophrenia diagnosis (mean age: 43 years, male: two, female: five). The verbal contents were transcribed, then two independent coders did the categorisation of the content; investigator triangulation ensured the reliability and validity of the study. As a result, a five-level hierarchy of cognitive operations was revealed in a bottom-up way. Patients attempted to interpret not only facial expressions but expressive movements, too, during emotion recognition. When deducing the state of others, patients sought contextual information; moreover, the burden of the illness and stigmatisation became easy to share. No signs of awareness of the consequences of the inferred states were found regarding future adjustment. Our hierarchical structure was consistent with related literature, and the specificities of cognitive operations of participants were also revealed.