We compared three arterial line insertion techniques and two types of arterial catheters in 69 critically ill patients. Use of the direct-puncture technique (method A) was associated with a significantly higher failure rate (23%) than use of a catheter with a separate guide wire (method B, 'classical' Seldinger technique, p < 0.001) or a catheter with an integral guide wire (method C, 'modified' Seldinger technique, p < 0.02). Operators randomly allocated to using method A took significantly longer to perform the procedure than those using method C (p < 0.01), used significantly more catheters (p < 0.0001) and made significantly more punctures in achieving a successful insertion than those using either methods B (p < 0.001) or C (p < 0.001). Both catheter types B and C (polyurethane) were significantly less likely to block, thus requiring less likely to block, thus requiring re-insertion, than catheter type A (Teflon) (p < 0.02, p < 0.01 respectively). We recommend the use of a 'classical' Seldinger technique (method B) for arterial line insertion in critically ill patients and the use of a polyurethane catheter, in preference to Teflon, to maximise catheter life after insertion.