For rating the noise of broad‐band sources the custom is to utilizeA‐weighted levels, but there is no consensus on the subjective importance of pure‐tone components that may be present. This study was to provide some direct evidence on this point. Nine subjects participated, individually, and the procedure was to have the subject match pairs of noises by adjusting the level of one of them. Two experiments used a modified random noise to which a pure tone in the range 250–2000 Hz could be added. The first experiment involved a simple comparison in a quiet environment. The second involved similar noises but with the addition of a local “wanted” sound in the form of music, and the subject matched noises from the viewpoint of “intrusiveness.” A third experiment was similar to the first but utilized actual machine noises as the basic spectra. All three experiments indicated that theA‐weighted level remains an adequate (slightly conservative) rating number even when pure tones are present.