Protein glycation and formation of advanced glycation end products (AGEs) play an important role in the pathogenesis of diabetes mellitus (DM) complications, neurodegenerations and age-related diseases. Prediction model for antiglycation activity can reduce costs and increase productivity and quality of screening investigations. Azolo[5,1-c][1,2,4]triazines and azolo[1,5-a]pyrimidines are well known biologically active compounds, which additionally have antiglycation properties. Hence, a number of 4-hydroxy-4H-azolo-1,4-dihydro [5,1-s]-1,2,4-triazines were selected for the prediction model creating. It has been established that azolotriazine derivatives have an anti-glycation effect, inhibiting a glycation of bovine serum albumin (BSA) by glucose with equal or greater activity than aminoguanidine. The activity range at 1000 μM concentration for variously substituted derivatives is 23.0–71.6% (30.3 ± 1.2% for aminoguanidine). The highest activity is detected for (4-hydroxy-4H-3-cyano-triazolo-1,4-dihydro[5,1-s]-1,2,4-triazines). The levels of antiglycation activity for the compounds (excluding aminoguanidine) correlate with the magnitude of the values of difference between HOMO and LUMO energies (∆_(HOMO-LUMO), (HOMO – highest occupied molecular orbital, LUMO – lowest unoccupied molecular orbital), established by PM3 semi-empirical method. Using the method of artificial neural network modeling, a mathematical model for describing the dependence of antiglycation activity on the calculated energies is obtained. It has been established that the E_LUMO and ∆_(HOMO-LUMO) energies have the largest contribution to the activity. The model can be used for the prediction of antiglycation activity.