The relationship between diesel exhaust odor and build and setting of the engine is reported. Odor was measured by a panel of observers, using a differential technique. Two odors were presented in each test, one from a reference engine, and one from an engine with variable characteristics. The observers were asked to score the difference between the two odors. This technique reduced the effect of random variations in the observers' sensitivity.The exhaust was passed through a coarse wire-wool filter before presentation to the observers. Thus the results apply to virtually soot-free exhaust. The observers found that odor decreased as load increased, but increased as speed increased. The catalyst removed a substantial amount of odor, but the remaining variables had a far smaller effect.