Eighty-seven patients, aged 15-84 years (mean, 61.8 +/- 16.2 SD), with essential tremor wrote two series of cursive e's and cursive l's on a standard sheet of ruled paper that was mounted on a commercially available digitizing tablet. Forty patients also drew an Archimedes spiral. Postural wrist tremor was measured with a triaxial accelerometer on the dorsum of the horizontally extended hand. The digitizing tablet was sufficiently sensitive to measure sustained visible tremor. Very severe tremor could not be recorded when it prevented a patient from keeping the ballpoint pen on the tablet. The intertrial variability of the handwriting data was such that a 36.0% change in mean acceleration amplitude (cm/s2) and an 8.3% change in mean tremor frequency (Hz) could be detected in the hypothetical population of 30 patients (paired-sample t test, p = 0.01, power = 90%). The intertrial changes detectable with accelerometry were 35.9% (amplitude) and 7.8% (frequency). The correlations between wrist tremor and writing tremor were < 0.60 for amplitude and < 0.25 for frequency. Significant correlations between patient age and tremor frequency and between tremor amplitude and frequency existed for postural tremor but not for writing or drawing tremor. Standard digitizing tablets for personal computers are useful in the quantitative assessment of writing tremor. The amplitude and frequency characteristics of tremor in posture, writing, and drawing may differ significantly.