Objective: to improve the efficiency of differential diagnosis of chronic pulmonary aspergillosis (СPA) based on the assessment of its probability using a discriminant mathematical model. Material and methods. The prospective study included 74 patients with CPA (57% women, median age 53 years) meeting the ERS/ESCMID criteria (2016). The control group consisted of 35 patients with lung diseases without CPA. Clinical and anamnestic data, the results of computed tomography (CT), laboratory and instrumental methods of research were analysed. By means of stepwise discriminant analysis, the model was created in order to differentiate compared groups. Results. The main forms of CPA were simple solitary aspergilloma (n = 30, 40%) and cavitary CPA (n = 21, 28%). On CT scans, in patients with CPA pulmonary emphysema (n = 50, 74%; 95% CI 63–83), bronchiectasis (n = 42, 56%; 95% CI 44–67), pleura thickening (n = 40, 56%; 95% CI 42–65) were detected with a high frequency. The sensitivity and specificity of typical for CPA air sickle symptom were 66.2% and 74.29%, respectively. The diagnostic informativeness of laboratory methods was characterized by high specificity (85–100%), however, it had sensitivity 40–60%. A discriminant model was worked up. It included five variables: mycological confirmation of the diagnosis (р < 0.001), air sickle symptom on CT (p = 0.03), ground glass opacity sympton on CT (p = 0.017), accompanying rheumatological diseases (p = 0,031), positive Aspergillus antigen in bronchoalveolar lavage (p = 0.036). The resulting model of differential diagnosis is statistically significant (F = (5.102) = 27.291; p < 0.001). Conclusion. CT-patterns of CPA include typical (air sickle symptom) and nonspecific (pleura thickening, emphysema, bronchiectasis) changes. Separately taken laboratory indicators and CT-symptoms are not always the determining criteria for diagnosis; an integrated approach is required to make a diagnosis. The proposed model improves the accuracy of differential diagnosis between CPA and nonmycotic lung diseases: increases sensitivity to 82.43%, specificity to 94.28% in comparison with separately analyzed laboratory data and typical CT-pattern of air sickle symptom. As a whole this model allows to classify the CPA and nonmycotic lung disease in 86,23% of cases.