A study of the sources of variability in particulate measurements using the Heavy-Duty Transient Test (40 CFR Subpart N) has been conducted. It consisted of several phases: a critical examination of the test procedures, visits to representative facilities to compare and contrast facility designs and test procedures, and development of a simplified model of the systems and procedures used for the Heavy-Duty Transient Test. Some of the sources of variability include; thermophoretic deposition of particulate matter onto walls of the sampling system followed by subsequent reentrainment in an unpredictable manner, the influence of dilution and cooling upon the soluble organic fraction, inconsistency among laboratories in the engine and dynamometer control strategies, and errors in measurements of flows into and out of the secondary dilution tunnel. The paper presents an examination of these and other sources of variability, gives recommendations for decreasing variability, and identifies critical research areas. Principal recommendations include: standardized engine and tunnel conditioning procedures, reduction of heat transfer during sampling and dilution, control of dilution ratio, tighter specification of the test cycle, and more accurate measurement of flows to and from the secondary dilution tunnel. Details of particle dynamics which were unimportant when particulate emission standards were greater than 0.1 gm/bhp-hr must be re-examined.