In this, the first report of precision and accuracy in simulator studies, ceramic-ceramic implants with ultra-low wear trends represented a relevant wear model. The effect of test durations was examined in a standard simulator test mode on the quality of the linear regression trends, the average wear estimates, and the amount of noise in the data. Three sets of diametral tolerances were compared in 28 mm diameter alumina implants. The authors' hypothesis was that wear data would be significantly improved with increased test durations. The average wear rates varied little with test duration, the biggest change amounting to only 30 and 15 per cent decreases in the wear estimate by 10 and 14 million cycles respectively. The most satisfactory improvement in the study was the decrease in variance (noise) with increasing duration, +/- 200 per cent at 5 million cycles reduced to +/- 55 per cent at 14 million cycles. The quality of the linear regression coefficients improved 150 per cent by 10 million cycles and 250 per cent by 14 million cycles. Overall the ceramic implants with highest diametral tolerances showed the least wear (15 per cent less, but not statistically significant). However, given such low wear rates for alumina liners, it was unlikely that any differences owing to diametral tolerances would be clinically significant in the typical patient.