BACKGROUND : Measurements of organ flow and perfusion during cardiopulmonary bypass suggest that perfusion of the splanchnic bed can be impaired by non-pulsatile flow. We postulated that non-pulsatile flow from centrifugal ventricular assist devices might also compromise splanchnic blood flow and cause bowel ischaemia especially in the period of circulatory instability early post-implant. The aim of the present studies was to compare the incidence of gastrointestinal (GI) complications in patients having a non-pulsatile device with the incidence in those having a pulsatile device. METHODS : In a pilot study, the initial 12 patients who received the Ventrassist (Ventracor, Sydney, NSW, Australia) centrifugal, non-pulsatile device during the period from June 2003 to September 2005 at the Alfred Hospital, Melbourne were compared with 11 patients who received a Thoratec (Thoratec, Pleasanton, CA, USA), pulsatile, positive displacement device and the incidence was recorded of GI complications requiring an intervention either surgical, endoscopic or by interventional radiology. This was followed by a larger (full) study of a second cohort of similar ventricular assist device (VAD) patients from January 1992 until December 2012 comparing 53 patients having non-pulsatile devices and 110 having pulsatile devices. RESULTS : In the pilot study, the overall incidence of complications in the non-pulsatile group (67%) was almost double that in the pulsatile group (36%) but the difference was not statistically significant (p = 0.15) because of the small number (n = 23) of participants. In the full study, all GI complications with either device occurred within the first 3 weeks post-implant. In the non-pulsatile patients, there was a higher incidence of GI bleeding, 23% vs 4% (p = 0.002), endoscopies, 24% vs 12% (p = 0.049). More patients with a non-pulsatile flow device had delayed absorption of nasogastric feeds than their pulsatile counterparts, 35% vs 7% (p < 0.0001). Patients with a non-pulsatile flow device had a higher overall rate of gastrointestinal complications than patients had with a pulsatile flow device, 56% vs 20% (p < 0.0001). After correcting for the other predictors, the odds of developing a gastrointestinal complication in the pulsatile group was significantly lower (odds ratio 0.07) than in the non-pulsatile device group (p < 0.0001). CONCLUSIONS : We conclude that the use of non-pulsatile centrifugal VADs compared with pulsatile positive displacement VADs is associated with a higher incidence of both haemorrhagic and ischaemic complications in the gastro-intestinal system especially in the very early post-implant period. Whether these complications could be reduced in centrifugal devices by increasing their pulsatility is not clear and merits further research.