Although the Centers for Disease Control recommends that needles should never be recapped, many phlebotomists routinely recap and change needles before blood culture inoculation. This study compared the extrinsic contamination rate in blood cultures when the needle was and was not changed. One hundred eight medical students obtained 182 blood specimens from each other by means of standard methods. Each specimen was inoculated into two culture bottles. The first bottle was inoculated with the needle used for phlebotomy, and the second was inoculated after needle change. Four (2.2%) of 182 bottles were contaminated when the needle was not changed, compared with one (0.6%) when the needle was changed. This small difference was not statistically significant, and the likelihood of having failed to detect a 5% difference in contamination rate was small. The risk of needle-stick injury incurred by changing the needle before inoculation of blood culture bottles seems to be unjustified.