Data from two test fleets, representative of the U.S. car population, were used to evaluate the relative merits of procedures using idle and short loaded emissions tests to reduce vehicle exhaust emissions. The incidence of engine malfunctions and the effect of each on idle emissions were determined. From this, a diagnostic procedure based on non-loaded testing (idle and 2500 RPM emissions) was developed. Tune-up repairs beyond those indicated by the procedure produced no significant further emissions reduction.The data show that non-loaded emissions testing combined with the non-loaded test diagnostic procedure can be as effective as loaded testing combined with a loaded test diagnostic procedure in reducing emissions both at idle and under load. The study also showed that the non-loaded test diagnostic procedure more correctly identified the malfunctions responsible for cars failing the emissions tests than did the loaded test procedure. In a number of cases this involved carburetor replacements that did not have to be made for cars to pass idle and loaded emissions tests. Nevertheless, analysis of EPA data suggests that such carburetor replacements would reduce emissions still further. Idle adjustments only (air-fuel mixture, idle speed, basic timing) were capable of repairing 55% of the rejected test cars so that they would pass emissions tests. Further repair of this group produced no significantly additional emissions reduction.