Spinal needles with a pencil-point tip and those of a finer gauge are known to be associated with a lower incidence of postdural puncture headache. This study set out to determine if fine pencil-point needles were acceptably easy to use in routine clinical practice. Two hundred and twelve women undergoing elective Caesarean section were randomly allocated to receive a subarachnoid block using either a 25 G or 27 G Whitacre needle. Factors determining ease of needle use, adequacy of block, incidence of postdural puncture headache, backache and neurological sequelae were assessed. Successful intrathecal injection was achieved in all patients in the 25 G group. Using the 27 G needle, the anaesthetist failed to reach the subarachnoid space in eight patients of which seven subsequently had a successful intrathecal injection with a larger needle. These failures were attributed to excessive needle flexibility which was the only significant difference in ease of use between the 25 G and 27 G needles. In the 25 G group, there was one severe postdural puncture headache which required an epidural blood patch and three mild headaches which resolved spontaneously. There were no postdural puncture headaches in the 27 G group. We conclude that the final choice of needle is a compromise between the ease of use and lower failure rate of the 25 G needle and the, as yet unproven, possibility of a lower incidence of postdural puncture headache with the 27 G needle.