Tissue irritation, as evidenced by phlebitis, associated with Optiva (Johnson & Johnson Medical) and Insyte (Becton Dickinson) polyurethane cannulae was studied. The integrity of the cannulae on removal, the incidence of infection at the cannula site and the factors which influence phlebitis were also examined. One thousand and eight patients had a polyurethane cannula placed for induction of anaesthesia for cardiac surgery. After surgery, the cannula was examined every 24 hours. If evidence of phlebitis occurred, the cannula was removed and sent for culture. All remaining cannulae were removed at 72 hours and the site examined daily for a further three days. There were 503 Optiva and 505 Insyte cannulae studied. The distributions between the two cannulae with respect to patient characteristics, gauge of cannula, number of attempts and difficulty of insertion, cannula site and anaesthetist inserting were similar. The early removal rate for both groups was 47%. Overall phlebitis rate with Optiva was 31% and Insyte 33%. This difference is not statistically significant. The cumulative phlebitis rate increased with time but did not differ between the two types of cannulae. Minor tip distortion or shaft kinking of the cannulae occurred in 16.2% of Optiva and 23.5% of Insyte. This difference is statistically significant and may relate to the slightly more acute taper at the Optiva cannula tip. Both cannulae were similar in clinical performance.