The purpose of the present study is to improve the on-street exhaust noise measurement technique for regulating vehicles that emit unacceptably large exhaust noise. The method under development uses racing operation of engines with a wide-open throttle on a standing vehicle. In contrast, the engine operation condition considered in the conventional on-street measurement technique considers operation at 75% of rated power followed by the throttle valve being released to its original position. The current regulation considers the maximum noise level during this engine operation condition. Accordingly, the engine is not loaded during testing. This loading condition provides unsatisfactory correlation between the exhaust noise level measured by the on-street method and the ISO vehicle acceleration noise under the wide-open throttle condition. The present paper reports the advantages of the proposed method as well as some important factors in the proper application of the newly proposed method. In the present study, we perform noise measurement along highways, a vehicle experiment in an anechoic chassis dynamometer, and computer simulations. The results revealed that the proposed on-street measurement method can be used to approximately determine the exhaust noise level during full acceleration for most vehicle types. In addition, the sensitivities of the measured noise levels for the testing conditions were clarified. The second item in this study is efficient engine speed measurement using exhaust sound signal only without electric wire harnessing. A survey of the measurement systems proposed by three different companies revealed that the measurement principles of two of these systems have potential application to on-street exhaust measurement. The last item investigated herein is better indices of exhaust noise evaluation. Loudness defined by ISO was found to be much better than the conventional dB(A) scale in terms of correlation with subjective evaluation results.