Using computer fuzzy-logical models based on empirical values of climatic characteristics (rainfall, temperature and humidity) of long-term observations (1955-2018) from meteorological stations in the Kabardino-Balkarian Republic (Nalchik, Baksan, Prokhladny and Terek) and crop yields (winter wheat, spring wheat, corn, sunflower, millet, oats), dependence of crop yields on variations of climatic factors were analyzed and a specific forecast was given. Setting expected values of climatic characteristics in computer model, we received possible values of productivity for the next season. Uniformity assessment (Dixon and Smirnov - Grabbsas criterion), stability (Student and Fischers criterion), statistical importance of parameters of distribution and accidental errors were determined. Originality of the method is in the fact that in the form of input parameters of the model predictors, the previously calculated forecast values of the meteorological parameters for the next agricultural year were used, and at the output, the predicted values of crop productivity were obtained as predictants. Furthermore, recommendations on adoption of management decisions were developed.