The need for quality assurance (QA) programmes covering near-patient testing technologies has been acknowledged since the mid-1980s. However, most effort has been directed towards capillary blood glucose monitoring techniques. The most common biochemical near-patient testing device is the dry reagent strip or 'dipstick' for urinalysis. This report describes a quality assurance programme for urinalysis, used in the wards and clinics of a district general hospital, and in some of the general practitioner surgeries within its catchment area. The project describes the preparation of an aqueous 'urine' sample, the design of a report form, the dispatch of the sample and report forms to the ward/clinic/health centre, the receipt and scoring of the returned results, and the assessment of the results, both in terms of management information and sources of error. Samples were spiked to give a target value midway between two colour blocks for each analyte. Results were scored as +/- 1 if adjacent colour block to the target, +/- 2 for results two colour blocks (error) and +/- 3 (gross error) for results three or more colour blocks from the target value. For urine glucose, results showed an overall 14.7% error and 2.6% gross error rate. Similar scoring gave the following error and gross error rates for the other analytes: bilirubin (1.0%, 3.3%); ketone (4.3%, 0.3%); specific gravity (13.4%, 3.1%); pH (11.2%, 6.5%); blood (7.7%, 2.9%); protein (9.7%, 2.3%); and nitrite (gross errors 4.9%). Investigation of the results of the programme showed four types of error in dipstick testing-timing, misalignment, misunderstanding and transcription. Analysis of the results showed a reduction in error rates when an electronic reader was used (errors 2.0%, gross errors 0.75%), compared to reading against the colour blocks on the side of the bottle (7.7%, 1.6%) or using the colour blocks on a flat card reader (7.4%, 1.7%). The report demonstrates the importance of a QA scheme in providing a reliable screening service. It gives confidence to the users of dipsticks, information on training needs to clinical managers, and an understanding of both types of error and how they can be reduced.