An error model of the Constant Volume Sampler (CVS) and the exhaust emission analyzers using standard error propagation techniques has been developed. The model, based on the individual errors of the different equipment involved in the measurement process, was used to compute the total uncertainty in grams/mile of the regulated pollutants Non Methane Hydrocarbons (NMHC), Oxides of Nitrogen (NOx) and Carbon Monoxide (CO) emitted by a vehicle during the U.S. Federal Test Procedure. The uncertainties resulting from the errors introduced by the vehicle, the driver and the chassis dynamometer are not accounted for in this analysis.At Ultra Low Emission Vehicle (ULEV) standards, the analysis showed that errors associated with the measurement of Total Hydrocarbon (THC), NMHC and NOx resulted in a high level of uncertainty. However, the results showed that the CO measurement uncertainty is acceptable.The model showed that one of the major sources of error is the analyzer repeatability. A minimum of 19% uncertainty in NMHC at ULEV levels (0.04 grams/mile) can be expected using current instrumentation. The model also revealed that there would not be significant differences in uncertainty in testing a large or small vehicle for a given emissions standard.The model also showed that a Variable CVS system reduced NMHC uncertainties from 19% to 9.4% and NOx uncertainties, from 8% to 5% at ULEV levels. The model was also used to predict the potential benefit of using a direct measurement of NMHC. This error model could be utilized to specify the instrumentation sensitivity that will be required at ULEV standards.