quality of all analytical measurements rests ultimately upon the quality of reference materials This statement is especially appropriate in the determination of trace contaminant residues in environmental matrices. To support the need for a certified, quality-controlled, common database in the analysis of pollutant chemical agents in environmental substrates, the U.S. EPA's Environmental Monitoring Systems Laboratory (Las Vegas) and Environmental Monitoring and Support Laboratory (Cincinnati) jointly maintain several repositories of analytical grade reference materials under the Agency's Quality Assurance Reference Materials (QARM) Project. Operated under contract by Northrop Services in Research Triangle Park, NC, the Project currently offers standards of some 2700 compounds of environmental concern, including pesticides and their metabolites and degradation products; PCBs, chloroand bromodibenzodioxins and furans, and other halogenated organics; plasticizers; nitrosamines; polynuclear aromatics; and heavy metals and metalloids. Included among these are the Agency-identified compounds commonly known as the priority pollutants, those materials regulated under Appendix VIII (RCRA) and CERCLA (Superfund) legislation, and groundwater monitoring compounds. The program is currently being supplied by nearly 200 chemical manufacturing companies and 33 chemical supply houses. Additional commercially-unavailable compounds, especially select environmental reaction and degradation products, are synthesized and purified in-house. Standards are offered, without charge, to any and all laboratories worldwide engaged in environmental monitoring, research, and/or regulatory and enforcement activities. During 1986, more than 150,000 standards were distributed in response to 7,000 requests. The analyst using Project standards expects the highest quality materials at accurate purity values and concentrations (for solutions), often without knowledge of how the integrity of each is established and maintained. Prior to distribution, each compound is subjected to a rigorous quality assurance program, including (1) component identification, (2) purity assay of neat materials, (3) concentration verification of solutions, and (4) stability studies, using a variety of instrumental methods. Although acquired materials typically include a percent purity value, each is individually reanalyzed in a series of steps. Chemical identity is verified by mass spectrometry, introduced via gas chromatograph or alternately, by direct insertion probe. For select compounds, other spectroscopic methods-IR, NMR, and ultraviolet spectroscopy-are supplementally employed. Subsequently, chemical purity is assayed by (at least) two methods, normally one chromatographic (GC, HPLC, TLC) and one non-chromatographic (DSC, NMR, IR, etc.). Establishing purity of a neat material is often complicated by the lack of a uniform response for all components to a single mode of detection. Electron capture, flame ionization, nitrogen-phosphorus, Hall, or mass spectrometric detection are the chromatographic methods of choice for most compounds, whereas HPLC or TLC are used for highly polar or thermo-labile materials. Purity is established by the ratio of peak area of the compound of interest to the combined peak areas for all eluting components. Implicit in this procedure is the assumption that all components elute and produce equal response at the detector. For FID or MS (positive electron impact), the assumption of equal responses is reasonably accurate, although some differences in relative response are inevitable. Non-chromatographic methods of purity analysis include ultraviolet spectroscopy, NMR, melting point, and differential scanning calorimetry (DSC). The latter method is frequently used for highpurity compounds because the method is typically reliable at purities >97% and useful for a wide variety of compounds. In DSC analysis, purity is estimated using a van't Hoff plot of melting point data. Typically, agreement of results among methods is within 1-2% and an average value is reported. However, in some cases, a substantial discrepancy in results is observed, necessitating further analyses.