One of the most common diseases at high risk is arterial hypertension. A study was conducted on the results of expeditions to settlements in Yamalo-Nenets Autonomous Okrug. A total of 505 people were examined. The patients underwent the collection of clinical and instrumental data (blood pressure, cardiointervalography, bioelectrical impedance analysis, the determination of the threshold of taste sensitivity to salt), the results of which were included in the calculation. Nonlinear logit regression was used to construct models of arterial hypertension risk. According to the results of multivariate logistic data analysis, the equation was obtained, which included: diastolic blood pressure (OR= 1.4), the nature of working activity (OR= 1.4), adaptation index for Baevsky (OR= 1.11), taste sensitivity to salt (OR= 1.11), conclusion according to cardiointervalography (OR= 1.11), as well as age, intracellular fluid and duration of work in the North. The quality of predictive models is in sensitivity 83.5% and specificity of 93.1%, with a diagnostic accuracy of 90.0% (χ2=302.4; p