In pathogenicity assessment, RNA-based analyses are important for the correct classification of variants, and require gene-specific cut-offs for allelic representation and alternative/aberrant splicing. Beside this, the diagnostic yield of RNA-based techniques capable to detect aberrant splicing or allelic loss due to intronic/regulatory variants has to be elaborated. We established a cDNA analysis for full-length transcripts (FLT) of the four DNA mismatch repair (MMR) genes to investigate the splicing pattern and transcript integrity with active/inhibited nonsense-mediated mRNA-decay (NMD). Validation was based on results from normal controls, samples with premature termination codons (PTC), samples with splice-site defects (SSD), and samples with pathogenic putative missense variants. The method was applied to patients with variants of uncertain significance (VUS) or unexplained immunohistochemical MMR deficiency. We categorized the allelic representation into biallelic (50 ± 10%) or allelic loss (≤10%), and >10% and <40% as unclear. We defined isoforms up to 10% and exon-specific exceptions as alternative splicing, set the cut-off for SSD in cDNA + P to 30-50%, and regard >10% and <30% as unclear. FLT cDNA analyses designated 16% of all putative missense variants and 12% of VUS as SSD, detected MMR-defects in 19% of the unsolved patients, and re-classified >30% of VUS. Our method allows a standardized, systematic cDNA analysis of the MMR FLTs to assess the pathogenicity mechanism of VUS on RNA level, which will gain relevance for precision medicine and gene therapy. Diagnostic accuracy will be enhanced by detecting MMR defects in hitherto unsolved patients. The data generated will help to calibrate a high-throughput NGS-based mRNA-analysis and optimize prediction programs.