There has been considerable interest in using safeguarded needles to reduce needlestick injury. In a randomised design, we studied the efficacy and safety of two such needles (the Insyte AutoGuard and the Protective Acuvance), by comparing them with a conventional catheter needle (Insyte), for intravenous cannulation (18 G) in 150 patients and for intra-arterial cannulation in another 150 patients (20 G). For intravenous cannulation, the success rates were similar in the three groups but insertion of the AutoGuard or Acuvance catheter was significantly more difficult than the conventional catheter. For the Acuvance, the back-flow of blood into the chamber was sometimes too slow. For intra-arterial cannulation, insertion of the AutoGuard was significantly more difficult than the other two devices, mainly because the backflow chamber of the AutoGuard was too short so that the chamber often filled with blood before cannulation. Insertion of the Acuvance was significantly more difficult than the conventional catheter. For both intravenous and intra-arterial insertion, handling of the withdrawn needle was judged significantly safer in the AutoGuard group than the other two groups, whereas there was no significant difference in the safety between the Acuvance and conventional groups. In five subjects from the AutoGuard group, blood splashed on retraction of the needle. Blood contamination during needle withdrawal occurred frequently in the control and Acuvance groups, but rarely occurred in the AutoGuard group. Therefore, the AutoGuard needle is more suitable for intravenous cannulation, and the Acuvance is more suitable for intra-arterial cannulation.