The phlebotomy technique, particularly the use of small-bore needles, may influence the reliability of coagulation testing and platelet count. Routine coagulation tests were assayed in blood specimens collected from 22 consecutive patients in three separate, sequential phlebotomies, using butterfly devices with different needle sizes. Test results of samples collected with 23 and 25 G needles were compared with those obtained with the currently recommended 21 G needle. Although both the prothrombin time and activated partial thromboplastin time displayed a trend towards lower values employing the smaller 23 and 25 G needles, results did not differ significantly from the reference 21 G needle specimen, with the exceptions of D-dimer (25 G versus 21 G needle, 186 +/- 70 versus 178 +/- 66/ml, P < 0.01) and platelet count (23 G versus 21 G needle, 246 +/- 55 versus 254 +/- 56 x 10(-3)/l, P < 0.01; 25 G versus 21 G needle, 240 +/- 55 versus 254 +/- 56 x 10(-3)/l, P < 0.01). None of the mean biases recorded for the parameters was clinically meaningful, nor did they exceed the current desirable analytical quality specifications for desirable bias. Results of the present investigation suggest that, when a proper technique is used and within certain limitations, butterfly devices with small-bore needles may be a reliable alternative to draw venous blood for platelet count and coagulation testing.