The presence of wild-type RAS alleles, as determined by genotyping codons 12, 13, 59, 61, 117, and 146, is a prerequisite for personalized anti-EGFR treatment of metastatic colorectal cancer (mCRC) patients. Here we describe analytical validation of in-house developed massively parallel sequencing technology (MPS) in comparison to the in vitro diagnostics (IVD) certified qPCR method. DNA extracted from FFPE samples from CRC patients (n=703) and reference standards (n=33) were tested for KRAS and NRAS mutations in 6 codons of exons 2, 3, and 4 using deep amplicon sequencing (DAS) on a MiSeq benchtop sequencer (Illumina). Two different amplicon lengths and two different library preparation methods (long-RAS and short-RAS) were tested in order to evaluate their impact on DAS performance. In parallel, identical tumor DNA was tested by the following IVD assays: therascreen KRAS RGQ PCR Kit (Qiagen), cobas® KRAS Mutation Test (Roche Diagnostics), and SNaPshot assay (Thermo Fisher Scientific). Both DAS assays detected all the mutations present in reference standards and external quality control samples, except for the artificially generated KRAS codon 146 mutation. The DAS assays performed sufficient analytical specificity and sensitivity (≥0.95). The use of shorter amplicons prolonged the preparation steps but significantly improved the sequencing success rate of FFPE-derived DNA. RAS mutation frequencies in the Czech CRC patients were similar to previous reports, although rare mutations were also detected. DAS with short amplicons is a good strategy for routine assessment of somatic mutations in low-quality FFPE-derived DNA.