The currently available instrument for electrical detection of occlusal caries lesions {[}Electronic Caries Monitor (ECM)] uses a site-specific measurement with co-axial air drying. The reproducibility of this method has been reported to be fair to good. It was noticed that the measurement variation of this technique appeared to be non-random. It was the aim of this study to analyse how such a non-random reproducibility pattern arises and whether it could be observed for other operators and ECM models. Analysis of hypothetical measurement pairs showed that the pattern was related to measurements at the high and low end of the measurement range for the instrument. Data sets supplied by other researchers to a varying degree showed signs of a similar non-random pattern. These data sets were acquired at different locations, by different operators and using 3 different ECM models. The frequency distribution of measurements in all cases showed a single or double end-peaked distribution shape. It was concluded that the pattern was a general feature of the measurement method. It was tentatively attributed to several characteristics such as a high value censoring, insufficient probe contact and unpredictable probe contact. A different measurement technique, with an improved probe contact, appears to be advisable. Copyright (C) 2005 S. Karger AG, Basel.