BACKGROUND: The difficulty in advancing the epidural catheters to hoped-for levels after successful demonstration of loss of resistance in the epidural space has well been known. This study was undertaken to determine the optimal distance that a catheter could be threaded into the epidural space without coiling. METHODS: An epidural catheter (single orifice, end hole) was inserted at L2-3 or L3-4 interspace with bevel of the needle directed either cephalad (n=15) or caudad (n=25). After injection of 1 ml of iohexol dye through the catheter, fluoroscopy was taken to determine the position of catheter tip. The findings of 40 epidurographies were analysed. RESULTS: The lengths of catheters threaded into the epidural spaces without coiling were 3.0+/-1.3 cm (mean+/-SD) and varied from 1.0 to 8.0 cm. Only 10 per cent of the catheters threaded without coiling 5 cm beyond the intervertebral space of insertion. No statistical significance was seen between the length threaded without coiling for catheters in the cephalad direction (2.9+/-1.1 cm) and in the caudad direction (3.0+/-1.3 cm). Although the Tuohy needle was inserted using the midline approach with the intention of positioning the catheter in the midline, only in 34.8 per cent was the catheter indeed situated in the midline. Conclusion : This study shows the futility of attempting to thread a catheter more than 3 cm within the epidural space. In lumbar epidural anesthesia, it is desirable to insert a catheter at the nearest possible spinal level.