Background. The aim of chronic myeloid leukemia (CML) treatment with tyrosine kinase inhibitors (TKI) is not only effectiveness, but also safety. Long treatment duration makes the analysis of most significant complications very important. Pleural effusion (PE) is an important adverse event of dasatinib therapy with largely unclear cause. The optimal management of recurrent PE is unknown and the analysis of its treatment results is actual. Aim. To describe the characteristics of patients with recurrent PE at prolonged dasatinib treatment and to suggest the strategy of their management. Methods. Follow-up data of 23 CML late chronic phase patients at dasatinib therapy after imatinib failure in 2 clinical studies: phase II study comparing dasatinib 140 mg and imatinib 800 mg daily (N = 12) and phase III dasatinib dose-optimizing study in patients with imatinib-resistant or intolerant patients (N = 11). M:F ratio was 7:16. Median age at the beginning of dasatinib - 48 years (26-68), median CML duration - 11 years (4,1-19,2). The reason for TKI change was imatinib resistance: cytogenetic (N = 17) and hematological (N = 6). Results. Median duration of dasatinib treatment - 40 months (10-107); 13 patients (56,3%) are alive, 10 patients (43,5%) have died because of progression of CML. In 19 patients (82,6%) dasatinib treatment was stopped because of: blastic transformation - 6 (26,1%), hematological resistance - 3 (13,1%), cytogenetic resistance - 5 (21,7%), PE - 5 (21,7%). Four patients are still on dasatinib treatment with median duration 8,8 years (8,7-8,9), 3 of them retain complete/major molecular response. The best responses were: complete hematological response in 21 (91,3%), complete cytogenetic response - in 8 (34,8%), major molecular response - in 6 (26,1%) patients. Overall 8-year survival was 55,1%, progression-free survival - 55,4%, event-free survival - 26,1%. PE was observed in 11 (47,8%) patients, in 8 of them (72,7%) - recurrent. In one patient the prolonged PE was associated with fibrosis of adjacent lung and pleura. Median time to PE was 34 (6-83) months. In cases of PE dasatinib was interrupted (mean duration 21 d) and diuretics were started. Six patients (54,5%) also received corticosteroids. Five patients (45,6%) were treated with thoracocentesis. In recurrent PE the dasatinib dosage was decreased. The dasatinib discontinuation in 4 patients with recurrent PE has led to loss of major molecular response in 2 of them; in other 2 it is retained for 6,5 and 1,5 years. Event-free 8-year survival was 36,4% in patients with PE, 16,7% - without it. Discussion. The response rate in patients with PE was not worse, than in those without it. Most often PE begins at 3rd year of treatment; later events were only relapses. Among risk factors of PE 2 patients had arterial hypertension and hypercholesterinemia, 3 patients were > 65 yrs. The significantly high PE on rate (48%) was linked to high initial dasatinib dosage (> 100 mg/d) and bid prescription. We have not observed cases of absolute lymphocytosis due to large granular lymphocytes proliferation. The continuation of treatment generally leads to recurrences of effusion. According to our experience, once arising PE tends to recur. The compensation can be maintained with continuous treatment with diuretics. The prolonged PE may lead to fibrosis of adjacent lung and pleura. Prolonged treatment interruptions and decreased doses can cause treatment failure. The role of corticosteroids is unclear. Conclusion. Our experience in recurrent PE management at dasatinib treatment allow to recommend the usage of alternative TKI in patients with poor treatment response, and discontinuation of treatment in patients with deep molecular response with close monitoring of residual disease by PCR. Disclosures Turkina: Novartis International AG: Consultancy; Pfizer: Consultancy; Bristol-Myers Squibb: Consultancy.