Desk-top analysers that are simple to use, portable and free from technical requirements are now widely used for blood cholesterol measurement outside the traditional pathology laboratory setting. Test accuracy and reliability are essential if these portable desk-top analysers are to be of value in the assessment and management of elevated blood cholesterol. To determine their accuracy and reliability we compared the results obtained from four different desk-top analysers with those from a teaching hospital's routine laboratory method. The desk-top analysers assessed were the Ames Minilab, the Kodak DT60, the Boehringer Reflotron and the Abbott Vision. The precision of each desk-top analyser was within acceptable limits, defined as a coefficient of variation less than 5 per cent. The results from the Vision, Reflotron and DT60 related closely to those from routine laboratory method, with least squares linear regression slopes ranging from 0.92 to 1.06 and intra-class correlation coefficients from 0.95 to 0.99. The Minilab showed least agreement with the routine laboratory method and caution should be taken in the interpretation of cholesterol estimations made with this device.