Mutation-based molecular diagnostics of autosomal dominant polycystic kidney disease (ADPKD) is complicated by genetic and allelic heterogeneity, large multi-exon genes, duplication of PKD1, and a high level of unclassified variants (UCV). Present mutation detection levels are 60 to 70%, and PKD1 and PKD2 UCV have not been systematically classified. This study analyzed the uniquely characterized Consortium for Radiologic Imaging Study of PKD (CRISP) ADPKD population by molecular analysis. A cohort of 202 probands was screened by denaturing HPLC, followed by direct sequencing using a clinical test of 121 with no definite mutation (plus controls). A subset was also screened for larger deletions, and reverse transcription-PCR was used to test abnormal splicing. Definite mutations were identified in 127 (62.9%) probands, and all UCV were assessed for their potential pathogenicity. The Grantham Matrix Score was used to score the significance of the substitution and the conservation of the residue in orthologs and defined domains. The likelihood for aberrant splicing and contextual information about the UCV within the patient (including segregation analysis) was used in combination to define a variant score. From this analysis, 44 missense plus two atypical splicing and seven small in-frame changes were defined as probably pathogenic and assigned to a mutation group. Mutations were thus defined in 180 (89.1%) probands: 153 (85.0%) PKD1 and 27 (15.0%) PKD2. The majority were unique to a single family, but recurrent mutations accounted for 30.0% of the total. A total of 190 polymorphic variants were identified in PKD1 (average of 10.1 per patient) and eight in PKD2. Although nondefinite mutation data must be treated with care in the clinical setting, this study shows the potential for molecular diagnostics in ADPKD that is likely to become increasingly important as therapies become available.