Background: Hydroxyurea (HU) treatment of patients with essential thrombocythemia (ET), polycythemia vera (PV), and primary myelofibrosis (PMF) (MPNs) normalizes elevated blood cell counts within weeks in the large majority of patients. Studies on the impact of HU upon the kinetics of the JAK2V617F allele burden, leukocyte, and platelet counts over time are scarce. Purpose: Using data-driven analysis as a novel tool to model the kinetics of the JAK2V617F allele burden and blood cell counts over time during treatment with HU. Material and methods: Using serial measurements of JAK2V617F and correlation analysis of routine hematological values (the Hb-concentration, leukocyte count, platelet count, and lactic dehydrogenase), we present a detailed description and analysis of the kinetics of the JAK2V617F, leukocyte, and platelet counts and lactic dehydrogenase in 27 patients (PV = 18; ET = 7; PMF = 2), who were followed in the Danish randomized trial (DALIAH). To further analyze the JAK2V617F kinetics, we use a machine learning clustering algorithm to group the response patterns. Results: Response patterns were highly heterogeneous, with clustering resulting in 3 groups and 3 outliers. In the large majority of patients, HU treatment was initially associated with a modest decline in the JAK2V617F allele burden in concert with a decline in leukocyte and platelet counts. However, HU did not induce a sustained and continuous decrease in the JAK2V617F allele burden. Conclusion: Using data-driven analysis of the JAK2V617F allele burden, leukocyte, and platelet kinetics during treatment with HU, we have shown that HU does not induce a sustained decrease in the JAK2V617F allele burden and neither induces sustained normalization of elevated cell counts in MPN patients. Our results may explain why MPN patients during treatment with HU still have a substantially increased risk of thrombosis.