The genes encoding the LDL receptor and apoB were screened for mutations associated with familial hypercholesterolemia (FH) in 408 patients referred to the Lipid Clinic in 1995-2003. The study aimed at testing the ability of three different sets of clinical criteria to predict the results of molecular genetic analysis, and secondly test whether population-based age- and sex-specific percentiles of LDL-cholesterol offer useful supplemental information in the selection of patients for molecular genetic analysis. The patients were retrospectively categorised according to Simon Broome Register Group criteria, Make Early Diagnosis to Prevent Early Death criteria (MEDPED) and the Dutch Lipid Clinic Network criteria, and the distribution of patients was compared to the results of the molecular genetic analysis. The study illustrates a classical dilemma. Mutation detection rates (and specificities) are high only if sensitivity is very low and vice versa: to find most mutation carriers, even patients with only possible FH must be examined by molecular genetic testing leading to mutation detection rates as low as 30-40%.