This study evaluated the influence of a desk-top analyser, used in general practice, on the mean number of blood tests per contact, prescribed or analysed by general practitioners. A randomized controlled trial was used. The mean number of blood tests per contact requested by two groups of GPs, before and after the introduction of a Reflotron in the intervention group was compared. Practitioners were assigned to the control group or to the intervention group by stratified randomization. Flemish GPs, known by the Flemish Institute of General Practice as interested in research were enrolled into the study. There were two registration periods of 8 weeks each. The weekly number of doctor-patient contacts and all requested or performed blood tests for each patient were registered. During the second period the members of the intervention group were asked to use the Reflotron, following their own judgement, and to register the total number of tests per contact, performed with it, on a special form. In the Reflotron group there was a slight increase in the median of the relative differences between the intervention and the base-line period (3%). In the control group the median of the relative differences decreased (-7%). The difference between both groups was not statistically significant (P = 0.17). In both groups the size and direction of the relative differences of the individual practices were very different. No statistically significant differences were found in any of the subgroups.