Background. Assessment of lung injury in oncohematological patients is a relevant problem, since the spectrum of pathological changes is wide and includes pulmonary infections, tumor cell infiltration, cardiogenic and non-cardiogenic pulmonary edema, bronchiolitis obliterans, interstitial pneumonitis, post-radiation and post-inflammatory pneumofibrosis, pulmonary vasculopathy and pleural effusion. At the moment there are no approved recommendations with criteria of differential diagnosis for these conditions, in particular, with differences between the most common therapy complication represented by pulmonary infections and poorly explored drug-induced toxic lesions.Aim. Identification of criteria for pneumotoxicity, allowing for differential diagnosis with pulmonary infections developing during chemotherapy, according to data routinely obtained in real clinical practice.Materials and methods. The study group included 38 patients with cytotoxic and autoimmune lung injury caused by specific therapy (group 1); the comparison group included 38 patients with infectious lesions receiving the same antitumor drugs (group 2). The data of the anamnesis, clinical course, instrumental studies and standard laboratory tests was studied retrospectively. For statistical analysis, the Mann–Whitney, χ2, Kruskal–Wallis tests were used. ROC analysis was performed to assess the sensitivity and specificity of various factors in relation to toxic damage.Results. Patients with lymphomas predominated in group of toxic lung injury (63 %). In patients who underwent allogeneic hematopoietic stem cells transplantation, toxic complications developed in the period from 35 to 1289 days, infectious – from 4 to 43 days. Statistically significant differences were obtained in the presence of a concomitant state of an altered immune response: 32 % of patients in the toxic lesion group versus 5 % in the infectious group had a history of allergy, and, in contrast to the infectious lesion group, in the toxic lesion group autoimmune diseases were detected. The main symptom in patients of the first group was shortness of breath, which was observed in 68 % of cases, of the second – an increased body temperature, observed in 92 % of cases; cough was also a common symptom – in 19 % and 13 % of patients respectively. In 58 % of patients of the second group, concomitant mucositis was detected, while in the first group this complication did not occur in any of them. The most common radiological pattern (71 % of cases in each group) was ground-glass opacities, in patients of the second group often combined with infiltrative changes and thickening of the bronchial walls (in 53 and 42 % of cases respectively). Among laboratory results, the largest differences between groups were observed in the leukocyte levels (with an average level of 2.5 . 109 / L in the infectious group versus 6 . 109 / L in the toxic group), eosinophils (with an average of 3.6 % in the toxic group versus 1.75 % in the infectious group), C-reactive protein (with an average level of 146.7 mg / L in the infectious group versus 52.4 mg / L in the toxic group), and creatinine (with an average of 0.085 mmol / L in the toxic group versus 0.071 mmol / L in the infectious group).Conclusion. The data obtained in this research indicates the value of taking an anamnesis and the importance of performing additional studies in patients with suspected drug-induced lung injury, as well as identifies risk groups. Based on the revealed differences, a scale for the differential diagnosis of drug-induced toxic and infectious lung damage, which includes the results of publicly available research methods, with high sensitivity and specificity, was proposed. Further research for more specific, but, at the same time, universal for various drugs, criteria for toxic lung damage is relevant.