The incidence of thrombophlebitis following the use of two types of cannulae, a new generation polyurethane cannula Biovalve PUR and a polyurethane hybrid co-polymer Vialon cannula, was compared in a randomised trial. Seventy patients scheduled for elective cardiac surgery were randomly assigned to have either a Biovalve cannula or a Vialon cannula inserted before induction of anaesthesia. Postoperatively, the cannulae were removed either as soon as signs of thrombophlebitis were observed or at the end of seven days, whichever was earlier. There was a significantly lower incidence of thrombophlebitis in the Biovalve PUR group (17%) compared to the Vialon group (60%) at the end of seven days (p < 0.001). None of the cannulae tip cultures resulted in the growth of micro-organisms. There was also a lower incidence of the need for re-siting with the Biovalve than with the Vialon cannulae.