The constituents of raw tobacco products are naturally variable, and product design, the manufacturing process and the evolving science of tobacco analytics may increase this variability. The call for certified tobacco reference materials and their potential application in product research and regulation indicates a need to understand this variability. The United States Food and Drug Administration awarded a Cooperative Agreement to the Center for Tobacco Reference Products at the University of Kentucky to produce and characterize four smokeless tobacco reference products including two snus, a moist snuff and a loose-leaf chewing tobacco. The statistical method for certification highlights the impact of measurement variability and laboratory differences on the certification process and therefore ultimately on the Certificate of Analysis. Characterization data from four ISO-accredited laboratories were used for the certification. Of the 14 parameters examined, acetaldehyde, formaldehyde, free nicotine, total nicotine, arsenic and benzo[a]pyrene exhibited larger than expected measurement variability, and crotonaldehyde was below the limit of quantitation for all samples. Graphics of the data and Cochran and Grubbs outlier analyses illustrate the types of variability experienced for the six constituents with large measurement variability, including consistent laboratory means with large between-lab variance and large repeatability standard deviations within a specific lab.