A test procedure has been developed, called the “Leak-Rate Procedure,” which makes it possible to measure directly the weight of gasoline vapor lost from a vehicle tank during refueling. Experimental data show that this test procedure can be used to measure the losses from random vehicles with an accuracy of a few percent. The procedure can be used with displacement-type vapor recovery systems and may be useable with some secondary recovery systems. Data are presented to illustrate the problem with “baseline” test procedures. It was concluded that it is impractical to use the baseline procedure for determining vapor emissions when testing random vehicles at service stations where it is not possible to control or measure all of the variables all of the time.