One hundred elderly male patients undergoing transurethral surgery were allocated randomly to receive spinal anaesthesia with either a 26 gauge Yale needle or a 24 gauge Sprotte needle. Patients were visited within 48 h by an investigator who was unaware of the needle type used and specific enquiry was made about any headache which was characteristic of dural puncture. We found evidence of a postdural puncture headache in 15 (30%) patients in whom a 26 gauge Yale needle was used. By comparison, only three patients (6%) in the Sprotte group reported a postdural puncture headache. This represents a highly significant (P < 0.005) reduction in the incidence of postdural puncture headache. The incidence of multiple attempts at dural puncture was also significantly (P < 0.05) reduced to 16% in the Sprotte group compared with 28% in the Yale group.