The current national (ASTM E-492) and international (ISO-R140/R717) impact method involves the use of a standard hammer machine for evaluating the impact noise isolation of floor/ceiling structure. The method has been severely criticized on the ground that ratings based on the test data correlate poorly with the subjective judgments of people listening to real-life impacts on the same floors. An alternative test method is proposed that uses a modified hammer machine whose internal impedance, intensity of impact and striking frequency simulate those of real footfalls; the new method involves reading the "peak" (i.e., short-term rms) impact sound pressure levels in the receiving room, replacing the current practice of reading long-term rms levels. No normalization for the sound absorption of the receiving room is required. Since the peak impact levels, particularly for floors with good impact isolation, are much higher than the long-term rms levels usually measured, the interference of background noise is much less for the new method. These proposed changes, based on recent studies, are expected to lead to better correlation between the test data and subjective judgments of floors. On an experimental basis, it is proposed that the range of measurement be extended down to 40 Hz, because of the certainty that low-frequency components are significant in the subjective evaluation of wood joist floors and the probability that peak levels are less susceptible to the problems of standing waves and lack of diffuseness than rms levels.