The power absorbed by 12 male subjects during exposure to vertical whole-body vibration at six magnitudes of random vibration (0·25, 0·5, 1·0, 1·5, 2·0 and 2·5 ms−2r.m.s.) has been measured in the laboratory. All subjects showed greatest absorbed power at about 5 Hz, but the frequency of this peak in the absorbed power reduced with increasing vibration magnitude. The total absorbed power increased approximately in proportion to the square of the acceleration magnitude: normalizing the absorbed power to the square of the r.m.s. vibration magnitude removed most of the differences, although the changes in resonance frequency were still evident. The frequency dependence of absorbed power at a constant magnitude of acceleration was approximated by a simple weighting having slopes of ±6 dB/octave either side of 5 Hz. Comparing the characteristics of this absorbed power weighting to standard frequency weightings showed substantial differences, especially at high frequencies. It is concluded that the differences from currently accepted frequency weightings are so great that the absorbed power is unlikely to yield good general predictions of the discomfort or risks of injury from whole-body vertical vibration.