On the basis of multivariate statistical analysis developed method for the separation of informative indicators of heart rate variability (HRV), which have diagnostic value for the objective assessment of operator's emotional reaction from watching emotive videos. The technique consists of two stages: 1) multivariate analysis of variance; 2) stepwise discriminate analysis with different strategies of feature selection. The total average classification accuracy 4 states of operators by 10 informative indices (two of which are indicators of the results of self -report and other indicators HRV) reached 95.7 %.