Objective To explore the demographic and clinical features of severe catatonic patients, comparing responders and non-responders to ECT in order to detect possible predictors of non-response. Methods This naturalistic study included 59 catatonic inpatients with a diagnosis of mood disorder according to DSM-IV-TR criteria. All patients were treated with bilateral ECT and evaluated before and after ECT course. The response to ECT was defined as a Clinical Global Impression (Improvement subscale) rating 1 ‘very much improved’ or 2 ‘much improved’. Clinical variables were compared between responders and non-responders; logistic regression was used to predict the probability of non-response, with regard to the symptoms presented by the patients. Results The response rate was 83.1%. Non-responders (n = 10) to ECT showed neurological comorbidities, treatments with dopamine agonists and anticholinergic drugs, waxy flexibility, and echophenomena more frequently than respondents (n = 49). Echophenomena resulted a significant predictor of non-response in the multivariate analysis. Conclusion In line with previous reports, ECT resulted effective in the vast majority of severe catatonic patients. The association between ECT resistant catatonia and neurological comorbidity, use of dopamine-agonist and anticholinergic medications is consistent with the hypothesis that ECT is more effective in ‘top-down’ than in ‘bottom-up’ variant of catatonia. Key points Catatonic symptoms are frequently associated with severe and psychotic mood disorders. Electroconvulsive therapy is effective in treating most forms of severe catatonia. Neurological comorbidity and the presence of ‘echopraxia/echolalia’ could represent predictors of non-response to ECT.