The physical characteristics of worksite impulse noise were defined along with the parameters which have been suggested for use in relating this noise to damage of the ear. The physiological and psychoacoustic phenomena which can be used to develop a definition of impulse noise were also considered. Measurements of wideband noise were taken at 14 manufacturing sites in the eastern United States, each producing a different impulsive noise signature. Two types of recordings were taken: short samples and long duration samples. Six different processes were selected from the 14 sites visited for further analysis. These six were taken from bottling, tube mill, stamping, punch press, stapler, and drop forge operations. Peak sound pressure levels of the impacts ranged from 108.8 decibels (dB) for bottling to 134.4dB for forging. No relation existed between the peak level and the duration of the impact. Thus neither sample could serve as a basis of classification; nor did peak level measurements provide adequate information on which to base hazard assessment. Frequently the use of crest factors, the ratio of peak to root mean square sound pressure, as an impulse descriptor has been considered. A fixed time window over which a statistic is calculated could be used as a basis for classification. Selecting this window on the basis of the production of temporary threshold shift in the ear provided a biological basis for the definition of impulsiveness. Classification of impulsiveness based on the sample kurtosis met the requirements of a generally applicable impulse definition.