We hypothesised that infants with ventricular dysfunction after cardiac surgery have impaired haemodynamic response to arginine–vasopressin therapy. We retrospectively reviewed the medical records of neonates and infants treated with arginine–vasopressin within 48 hours of corrective or palliative cardiac surgery who underwent echocardiographic assessment of ventricular function before initiation of therapy. Patients were classified as “responders” if their systolic blood pressure increased by ⩾10% without increase in catecholamine score or if it was maintained with decreased catecholamine score. Response was assessed 1 hour after maximum upward titration of arginine–vasopressin. A total of 36 children (15 neonates) were reviewed (17 male). The median (interquartile) age was 10.4 weeks (1.1–26.9), and the median weight was 4.3 kg (3.2–5.8). Diagnoses included single ventricle (eight), arch abnormalities (five), atrioventricular septal defect (four), double-outlet right ventricle (three), tetralogy of Fallot (three), and others (13). In all, 12 patients (33%) had ventricular dysfunction. Only 15 (42%) responded favourably according to our definition 1 hour after the “target” arginine–vasopressin dose was achieved. Ventricular dysfunction was not associated with poor response. The overall mortality was 25%, but mortality in patients with ventricular dysfunction was 42%. Favourable response was associated with shorter ICU stay (9.5 days versus 19.5 days, p=0.01). We conclude that arginine–vasopressin fails to increase blood pressure in ~50% of hypotensive children after cardiac surgery. The response rate does not increase with duration of therapy. Ventricular function does not predict haemodynamic response. The mortality in this group is very high. Prospective comparison of vasopressin with other vasoactive agents and/or inotropes is warranted.