DNA sequence analyses have become a major component in the diagnostic work-up of patients; however, limited consideration appears to be given to the possibility that reported results may in fact be wrong. Over the last four years we have carried out an External Quality Assessment scheme for mutation analysis in phenylketonuria. Each year, three DNA samples with previously characterized genotypes were mailed to participating laboratories. Indications for testing were either confirmation of diagnosis and prediction of disease severity, or carrier analysis. Each year there were several laboratories that failed to identify mutations because of methodological limitations. Of the participating laboratories that used comprehensive mutation detection methods, each year there was at least one that missed at least one mutation. Indeed, in the 2007 scheme almost 8% of reports from laboratories that used comprehensive mutation detection methods such as sequencing of all exons of the PAH gene contained incorrect genotypes. There were also serious deficiencies in the interpretation of genotype data: in the 2007 scheme, 6 out of 10 laboratories that obtained full genotyping marks for interpretation incurred a reduction of marks because information on the expected phenotype was missing or wrong. Several laboratories failed to appreciate the clinical relevance of a mutation associated with mild hyperphenylalaninaemia, which does not require treatment, and some discussed the option of prenatal diagnosis in the respective case. In conclusion, mutation analyses may be prone to errors and this demands careful interpretation of results in relation to clinical and biochemical findings.