INTRODUCTION: Gastric cancer is a frequent malignant disease with poor prognosis. Most patients undergo only palliative treatment. Chemotherapy is another alternative but its effect differs in individual patients. METHOD: This is retrospective study. We enrolled 54 patients (N=54) according to the inclusion criteria. We performed quantification of gene expression of selected genes and some microRNA from tumour tissue, which was used for the diagnosis. Statistical analysis of the data was performed. RESULTS: We demonstrated a predictive value of gene expression of thynidylate synthase in tumour tissue for a therapeutic effect of chemotherapy based on 5-Fluorouracil or Capecitabine. At the same time, we demonstrated a predictive value of miR181, miR150, mir192 and miR342 microRNA levels from the tumour tissue. In addition, we succeeded to demonstrate a predictive value of miR221, miR224, miR520 and miR375 microRNA levels for a therapeutic effect of chemotherapy based on platinum derivates. CONCLUSION: Thanks to the use of efficient therapy predictors, we can distinguish those patients who will profit from chemotherapy from patients where an effect cannot be expected. Thanks to personified oncology therapy the quality of life of some patients can be improved while reducing the costs of the therapy by avoiding inefficient chemotherapy. Only an early diagnosis of gastric cancer can reverse the adverse prognosis of patients with this disease. KEY WORDS: gastric cancer - microRNA - prognostic markers - predictive markers.