This study compared the effects of buffered versus unbuffered lidocaine when used as local anesthetic in preparation for cannulation of the arterial and venous sites of children requiring hemodialysis. The subjects for this study were seven children, ages 6-18 years, observed during 101 dialysis treatments. For each subject undergoing hemodialysis on a given day, one syringe containing 1% lidocaine (L) and one containing buffered lidocaine (BL) were prepared. The BL solution was prepared by adding 2 mL of 8.4% sodium bicarbonate to 20 mL of 1% lidocaine just prior to use. The choice of local anesthetic used for cannulation of the arterial or venous site was randomly assigned to be either L or BL. Nurses, raters, and subjects were blind to contents of the syringe. The procedures for piercing the skin, pausing, and infiltrating were standardized, as was the volume administered. Speed of injection was not controlled. Comparisons of self-reported pain and behavioral observations for L versus BL revealed no significant differences for pain of infiltration or pain of cannulation. Technique variables such as the speed of injection, which tended to be very slow for these children, appear to significantly influence infiltration pain and the relative merits of buffering.