The aim was to establish the prevalence of pulmonary embolism in 21 children (median age 12 months; range 5-132 months) with central venous lines in situ > 3 months (median 10 months; range 3-47). Twelve-lead electrocardiograms (ECGs) and echocardiograms were analysed in a retrospective study using ECG and echocardiographic criteria for pulmonary embolism-previously established and validated in adult patients- and standard paediatric ECG values as control data. Patients were scored as having definite (n = 7), probable (n = 5), or no pulmonary embolism (n = 9). Overall 57% of ECGs showed abnormalities compatible with pulmonary embolism. In two patients, serial ECGs obtained during an acute cardiorespiratory illness showed cumulative changes diagnostic of pulmonary embolism. Eight of 12 patients with abnormal ECGs had echocardiography; in seven of these (88%) the right ventricular end diastolic diameter was > 2SD above the mean value for age. Twelve of the patients included in this study have died; two died following an acute respiratory illness. There was postmortem evidence of pulmonary thromboembolism in both of the two children for whom necropsy information was available. The data suggest that pulmonary embolism is common in children who have central venous lines in situ for > 3 months. Serial studies are of value in some patients. Pulmonary embolism may compromise the long term survival of children with small bowel failure and preclude consideration for liver and small bowel transplantation.