The FHWA traffic noise prediction model (STAMINA) has been adopted in Ontario because of its flexibility and analytical features, which accommodate changed conditions through simple updating procedures. Major inputs for STAMINA are the reference energy mean emission levels of vehicle classes as a function of speed. These functions were established by the FHWA in their original report on the basis of data collected in the United States before 1978. However, conditions in Ontario in 1985 are different, and the noise emission level functions used in the STAMINA and other related programs should be reevaluated. Data on reference emission levels of cars and of medium and heavy trucks were collected during 1984 and 1985, processed, and statistically analyzed. From these data, functions of reference noise emission levels with vehicle speed were established for those vehicle groups. These functions can be used in programs derived from the FHWA model. The findings in Ontario confirm those in other jurisdictions in the United States, namely, that heavy trucks emit less noise at high speeds than originally indicated by the FHWA model. Further, it is shown that about 4 percent of heavy trucks are notoriously noisy compared with the general population and cause an upward shift of the reference emission level function by 0.5 to 1 dBA. These noisy trucks are relatively rare events, which may or may not be missed in noise measurements of short duration (20 min), but they have a high impact on the level of noise pollution.