Given q as the global frequency of the alleles causing a disease, any allele with a frequency higher than q minus the cumulative frequency of the previously known disease-causing mutations (threshold) cannot be the cause of that disease. This principle was applied to the analysis of cystic fibrosis transmembrane conductance regulator (CFTR) mutations in order to decide whether they are the cause of cystic fibrosis. A total of 191 DNA samples from random individuals from Italy, France, and Spain were investigated by DGGE (denaturing gradient gel electrophoresis) analysis of all the coding and proximal non-coding regions of the gene. The mutations detected by DGGE were identified by sequencing. The sample size was sufficient to select essentially all mutations with a frequency of at least 0.01. A total of 46 mutations was detected, 20 of which were missense mutations. Four new mutations were identified: 1341+28 C/T, 2082 C/T, L1096R, and I11131V. Thirteen mutations (125 G/C, 875+40 A/G, TTGAn, IVS8-6 5T, IVS8-6 9T, 1525-61 A/G, M470V, 2694 T/G, 3061-65 C/A, 4002 A/G, 4521 G/A, IVS8 TG10, IVS8 TG12) were classified as non-CF-causing alleles on the basis of their frequency. The remaining mutations have a cumulative frequency far exceeding q; therefore, most of them cannot be CF-causing mutations. This is the first random survey capable of detecting all the polymorphisms of the coding sequence of a gene.