Effective targeted therapy of pulmonary arterial hypertension (PAH) and chronic thromboembolic pulmonary hypertension (CTEPH) requires regular risk stratification. Among many prognostic parameters, three hemodynamic indices: right atrial pressure, cardiac index, and mixed venous saturation are considered critically important for correct risk classification. All of them are measured invasively and require right heart catheterization (RHC). The study was aimed to verify assumption that a model based on non-invasive parameters is able to predict hemodynamic profile described by the mentioned invasive indices. A group of 330 patients with pulmonary hypertension was used for the selection of the best predictors from the set of 17 functional, biochemical, and echocardiographic parameters. Multivariable logistic regression models for the prediction of low-risk and high-risk profiles were created. The cut-off points were determined and subsequent validation of the models was conducted prospectively on another group of 136 patients. The ROC curve analysis showed the very good discrimination power of the models (AUC 0.80–0.99) in the prediction of the hemodynamic profile in the total validation group and subgroups: PAH and CTEPH. The models indicated the risk profiles with moderate sensitivity (57–60%) and high specificity (87–93%). The method enables estimation of the hemodynamic indices when RHC cannot be performed.