The purpose of this study was to compare results from five commercial hair testing laboratories conducting workplace drug testing with regard to bias, precision, selectivity and decontamination efficiency. Nine blind hair specimens, including cocaine-positive drug user specimens (some contaminated with methamphetamine) and negative specimens contaminated with cocaine, were submitted in up to five replicates to five different laboratories. All laboratories correctly identified cocaine in all specimens from drug users. For an undamaged hair specimen from a cocaine user, within-laboratory Coefficients of Variation (CVs) of 5-22% (median 8%) were reported, showing that it is possible to produce a homogenous proficiency testing sample from drug user hair. Larger CVs were reported for specimens composed of blended hair (up to 29%) and curly/damaged hair (19-67%). Quantitative results appeared to be method-dependent, and the reported cocaine concentrations varied up to 5-fold between the laboratories, making interlaboratory comparisons difficult. All laboratories reported at least one positive result in specimens contaminated with cocaine powder, followed by sweat and shampoo treatments. Benzoylecgonine, norcocaine, cocaethylene and hydroxylated cocaine metabolites were all detected in cocaine powder-contaminated specimens. This indicates that current industry standards for analyzing and reporting positive cocaine results are not completely effective at identifying external contamination. Metabolite ratios between meta- or para-hydroxy-cocaine and cocaine were 6- and 10-fold lower in contaminated specimens compared to those observed in cocaine user specimens, supporting their potential use in distinguishing samples positive due to contamination and drug use.