Introduction. Bronchial asthma (BA) is a relevant social and medical problem in our country and around the world. Currently, phenotypes of the disease are distinguished. One of the original decisions in phenotypes distinguishing is the use of cluster analysis. However, the concept of BA phenotypes at the health resort period of rehabilitation has not yet been formed. Aim — to determine the BA phenotypes upon admission of patients to a health resort medical rehabilitation (HRMR) using cluster analysis and to offer personalized rehabilitation programs. Materials and methods. 518 patients with asthma who underwent HRMR on the southern coast of Crimea were examined. Each patient received clinical, functional and laboratory examination. HRMR included correction of long-term treatment according to the severity of asthma, climate therapy, respiratory therapy, educational programs, and physiotherapeutic procedures. We applied cluster analysis in order to identify BA phenotypes. Description statistics methods were used for phenotype-cluster characterization, comparative analysis — for determination of reliable phenotypic characteristics and relation of the effectiveness of HRMR and phenotypes. Results. A model of three phenotype-clusters was developed. The first cluster included patients with BA of moderate severity, uncontrolled course, frequent exacerbations, history of atopy, a tendency to obesity, moderately reduced external respiration function, fixed airway obstruction, high adherence to long-term therapy with medium doses of inhaled glucocorticoids (ICS) combined with long-acting β-2 agonists (LABA). Cluster 2 included patients with mild asthma, controlled or partially controlled course of the disease, with rare short exacerbations, late onset, preserved external respiration function and exercise tolerance, but low adherence to long-term therapy with medium and low doses of ICS. Cluster 3 included patients with moderate to severe BA, uncontrolled course, with early onset, frequent and prolonged exacerbations, severe symptoms, significantly reduced external respiration function with fixed obstruction, decreased exercise tolerance, but low adherence to long-term therapy (4th stage) with high doses of ICS in combination with LABA and long-acting anticholinergics. A close relationship was found between phenotypes-clusters and the achieved effects: a significant increase in the control of BA and a high efficiency of rehabilitation in patients of the 1st and especially 3rd clusters and low in the 2nd cluster. The optimal rehabilitation programs for each of the selected cluster phenotypes were determined. Conclusion. The cluster model developed with the help of artificial intelligence has demonstrated high prognostic value in the determination of the effectiveness and change of control over the course of asthma as a result of HRMR. Personalized HRMR programs are suggested.