A major problem in Vehicle Exhaust Emissions and Fuel Economy Testing has been the variability in the measurements. An extensive test program was undertaken to identify and quantify the sources of variation. The test program was designed using four test vehicles on five CTE-50 Clayton Dynamometer cells whereby four repeats for each combination were provided for a total of 80 CVS-C/H Emissions and F.E. Tests. During each test a total of 23 test variables were monitored and recorded in real time. Nine other variables were also observed for a total of 32 variables: 5 vehicle related variables, 16 engine related, 4 dynamometer variables, 3 driver related, and 4 environment variables.The paper reports various results of this experiment and describes an instrumentation package that was developed for “on-line” data recording. A brief discussion of the software used for “off-line” processing of the voluminous data as well as several statistical analysis techniques which were developed especially for identifying and quantifying the sources of variation, are included. The benefits which result from reduced test variability, namely lower test cost because of improved test efficiency and increased confidence in results, are pointed out.