Objective: Blood collection is a critical part of the preanalytical phase of laboratory testing. Only a few procedures are evaluable for detecting errors in this non-automatic activity. Information about potential sources of error is frequently absent from quality-control procedures and training materials. Objective: To evaluate the performance of phlebotomists and to identify the major sources of errors during diagnostic blood collection. Methods: We evaluated the performance of 3 phlebotomists each from 10 laboratories regarding tourniquet time, request for fist clenching, excessive friction during skin cleaning, sequence of vacuum-tube usage, and mixing of tube contents after specimen collection. The total number of these laboratories represented an equal number of private (ie, owned by private parties) and public (ie, administration by government organizations) settings. Results: An error rate of greater than 60% was observed in the performance of procedures we examined, and the error rate was not significantly different between public and private settings other than more adequate mixing of the contents of primary collection tubes in private facilities. Conclusions: Most facilities did not achieve adequate quality standards in phlebotomy. In procedural manuals, training materials and practices for phlebotomists, greater emphasis should be placed on presenting information on possible sources of error and on the correct implementation of procedures that ensure the quality of diagnostic blood specimens collected for laboratory analysis.