The aim is to analyze molecular abnormalities in patients with non-small cell lung cancer (NSCLC) using next-generation sequencing. The study included patients with the histologically and morphologically confirmed NSCLC: 57 patients were diagnosed with lung adenocarcinoma and 56 with squamous cell lung cancer. 76 tissue samples were frozen and 37 were formalin-fixed paraffin-embedded. Genetic research was carried out using the Targeted Sequencing by MiSeq (Illumina). We sequenced 212 amplicon regions in 48 cancer-related genes on 103 samples that had passed the quality control (QC) process. After all QC steps, 10 samples did not pass it due to their low quality and/or insufficient amount of reads. In total, 1983 coding mutations were discovered. Out of them, 105 were nonstop mutations, and 78 were located in low frequency variants. We recorded most mutated genes (with significant mutations) in the cohort with all the variant frequencies below a 5% threshold with the variant depth above 100 reads and the frequency of 1141 variants and 842 variants respectively. We used the Personal Cancer Genome Reporter (PCGR) to determine the clinical significance of mutations. Most highly frequent variants were located in the genes as follows: ATM, CDH1, FGFR2, IDH1, TP53, and SMAD4 , but they did not cover any clinically validated mutations (Tier 3). Only a few genes contained clinical mutations (Tier 1 – variants of strong clinical significance; Tier 2 – variants of potential clinical significance): IDH1, TP53, EGFR, PIK3CA, KIT, CTNNB1, STK11, NRAS, KRAS, and ABL1. 15 patients contained low frequency mutations in these genes, and 10 out of them were stage I patients without any lymphoid or metastatic infiltration. All of the identified mutations were predicted as damaging by most effective prediction algorithms. Rare mutations that occur, especially in cases of early stage cancer, may be difficult to capture, mostly due to its heterogeneity and low DNA quantities. They can be missed during standard sequencing and thus possibly lead to the possible recurrence of cancer after the treatment. This approach may provide a stronger incentive to determine the corresponding treatment course based on these low frequency variants.