Mutations to PKD1 and PKD2 are associated with autosomal dominant polycystic kidney disease (ADPKD). The absence of apparent PKD1/PKD2 linkage in five published European or North American families with ADPKD suggested a third locus, designated PKD3. Here we re-evaluated these families by updating clinical information, re-sampling where possible, and mutation screening for PKD1/PKD2. In the French-Canadian family, we identified PKD1: p.D3782_V3783insD, with misdiagnoses in two individuals and sample contamination explaining the lack of linkage. In the Portuguese family, PKD1: p.G3818A segregated with the disease in 10 individuals in three generations with likely misdiagnosis in one individual, sample contamination, and use of distant microsatellite markers explaining the linkage discrepancy. The mutation PKD2: c.213delC was found in the Bulgarian family, with linkage failure attributed to false positive diagnoses in two individuals. An affected son, but not the mother, in the Italian family had the nonsense mutation PKD1: p.R4228X, which appeared de novo in the son, with simple cysts probably explaining the mother's phenotype. No likely mutation was found in the Spanish family, but the phenotype was atypical with kidney atrophy in one case. Thus, re-analysis does not support the existence of a PKD3 in ADPKD. False positive diagnoses by ultrasound in all resolved families shows the value of mutation screening, but not linkage, to understand families with discrepant data.