An analysis has been performed to derive a frequency weighting for the development of vibration-induced white finger (VWF). It employs a model to compare health risks for pairs of population groups that are selected to have similar health outcomes from operating power tools or machines with markedly different acceleration spectra (rock drills, chain saws, pavement breakers and motorcycles). The model defines the Relative Risk, RR(f(trial)), which is constructed from the ratio of daily exposures and includes a trial frequency weighting that is applied to the acceleration spectra. The trial frequency weighting consists of a frequency-independent primary frequency range, and subordinate frequency ranges in which the response to vibration diminishes, with cut-off frequencies that are changed to influence the magnitude of RR(f(trial)). The frequency weighting so derived when RR(f(trial)) = 1 is similar to those obtained by other methods (W(hf), W(hT)). It consists of a frequency independent range from about 25 Hz to 500 Hz (-3 dB frequencies), with an amplitude cut-off rate of 12 dB/octave below 25 Hz and above 500 Hz. The range is compatible with studies of vasoconstriction in persons with VWF. The results provide further evidence that the ISO frequency weighting may be inappropriate for assessing the risk of developing VWF.