Within the framework of the European Concerted Action on Cystic Fibrosis (Biomed-2, BMH4-CT96-0462) a quality assessment was set up for 135 European and one Australian laboratory. Six DNA samples were sent to the various laboratories. These samples carried the following CFTR genotypes: dF508/N1303K; dI507/wild; dF508/G551D; dF508/621 + 1 GtoT; R553X/wild and 1717-1 GtoA/wild. Each laboratory was asked to process the samples as they routinely do, whether they checked for all mutations or not. More than 75% of the laboratories screened for at least six of these mutations. Heteroduplex analysis was the most frequently used primary testing method (47%), in many instances followed by restriction enzyme digestion. Only a minority of the laboratories made use of a commercial CFTR mutation detection kit. On average, 91% of the laboratories correctly typed both alleles of a given DNA sample. However, 35% of the laboratories incorrectly typed one or more alleles from a total of 12 alleles included in the trial. One laboratory even failed to identify four of the different alleles correctly. The genotyping error frequency tended to be lower in laboratories which perform more than 200 CFTR mutation analyses per year. The results of this quality control trial suggest that there are many laboratories (35%) which have a percentage of errors unacceptable in a routine testing setting. The development of a consensus testing strategy for routine diagnostic laboratories and centralised mutation analysis facilities for rare or country-specific mutations in a limited number of expert centres, in combination with regular training sessions and quality assessments, should further improve genotyping.