Objective To compare the ease of use of atraumatic needles with standard needles for diagnostic lumbar puncture and the incidence of headache after their use. Design Double blind, randomised controlled trial. Setting Investigation ward of a neurology unit in a university hospital. Participants 116 patients requiring elective diagnostic lumbar puncture. Interventions Standardised protocol for lumbar puncture with 20 gauge atraumatic or standard needles. Outcome measures The primary end point was intention to treat analysis of incidence of moderate to severe headache, assessed at one week by telephone interview. Secondary end points were incidence of headache at one week analysed by needle type, ease of use by operator according to a visual analogue scale, incidence of backache, and failure rate of puncture. Results Valid outcome data were available for 97 of 101 patients randomised. Baseline characteristics were matched except for higher body mass index in the standard needle group. By an intention to treat analysis the absolute risk of moderate to severe headache with atraumatic needles was reduced by 26% (95% confidence interval 6% to 45%) compared with standard needles, but there tvas a non-significantly greater absolute risk of multiple attempts at lumbar puncture (14%, – 4% to 32%), Higher body mass index was associated with an increased failure rate with atraumatic needles, but the reduced incidence of headache was maintained. The need for medical interventions was reduced by 20% (1% to 40%). Conclusions Atraumatic needles significantly reduced the incidence of moderate to severe headache and the need for medical interventions after diagnostic need for medical interventions after diagnostic lumbar punctures, but they were associated with a higher failure rate than standard needles.