Tremor rating scales (TRSs) are used commonly in the clinical assessment of tremor, but the relationship of a TRS to actual tremor amplitude has never been quantified. Consequently, the resolution of these scales is unknown, and the clinical significance of a 1-point change in TRS is uncertain. We therefore sought to determine the change in tremor amplitude that corresponds to a 1-point change in a typical 5-point TRS. Data from five laboratories were analysed, and 928 patients with various types of hand tremor were studied. Hand tremor was quantified with a graphics tablet in three different labs, an accelerometer in three labs and a mechanical-linkage device in one lab. Tremor in writing, drawing, horizontal posture, rest and finger-nose testing was graded using a variety of TRSs. The relationship between TRS scores and tremor amplitude was computed for each task and laboratory. A logarithmic relationship between a 5-point (0-4) TRS and tremor amplitude (T, measured in centimetres) was found in all five labs, despite widely varying rating scales and transducer methodology. Thus, T2/T1 = 10(alpha(TRS2-TRS1)). The value of alpha ranged from 0.414 to 0.441 for writing, 0.355-0.574 for spiral drawing, 0.441 to 0.488 for rest tremor, 0.266-0.577 for postural tremor and 0.306 for finger-nose testing. For alpha = 0.3, 0.4, 0.5, 0.6 and 0.7, the ratios T2/T1 for a 1-point decrease in TRS are 0.501, 0.398, 0.316, 0.251 and 0.200. Therefore, a 1-point change in TRS represents a substantial change in tremor amplitude. Knowledge of the relationship between TRS and precise measures of tremor is useful in interpreting the clinical significance of changes in TRS produced by disease or therapy.