Without control and complete specification of the experimental procedure, substantial differences can occur in the results of handle vibration measurements when a chain saw is hand-held and used to cut wood. At present, measurements on the same model saw by different laboratories can vary by over 200 in some octave bands, and similar variations can occur between production samples. Smaller variations, however, have been observed in the ISO frequency-weighted acceleration sum of the three components. In addition, the handle vibration of a new saw may not be a good indicator of the vibration exposure of a worker, owing to changes in vibration with engine speed, chain sharpness, isolator performance, deterioration of the saw with age and work techniques. The implications of these factors on the development of exposure risk criteria and regulations are discussed.