Introduction. We have examined 92 children aged between 6 and 15, suffering from chronic tonsillitis (CT). Tumor necrosis factor (TNF-α) and interleukin 1β and 6 (IL-1β and IL-6) contents have been defined in saliva. The control set comprised 17 healthy children. Cytokine content was defined with the enzyme multiplied immunoassay sets (Vektor Best Ltd., Russia) by enzyme-linked immunosorbent assay. The statistic analysis and data processing were carried out with statistic analysis programs (version 3.2, R Foundation for Statistical Computing, Vienna, Austria). Results. The content of cytokines TNF-α, IL-1, IL-6 in CT children’s saliva was high against the healthy children, yet the statistically significant differences were only noted for IL-6. In the CT group the median value of this factor (12.5) was significantly higher than in the control set (6.72) (p=0.01 in Mann—Whitney assessment). IL-6 was chosen as the basic factor for the mathematic model; its combinations in the form of a multi-factor logistic regression were given consideration. From out of the three possible models there was just one that had all the coefficients statistically significantly different from zero (TNF-α — IL-6). It was chosen as the basic diagnostic model for chronic tonsillitis. The created model’s sensitivity is 80.4%, while its specificity is 82.4%. Discussion. The revealed IL-6 dominance in saliva at CT can be is attributable to permanent antigenic challenge characteristic of the toxic allergic CT since, as previously shown, there are living proliferating microorganisms in the palatal tonsil tissues and their blood- and lymph vessels at CT. Conclusion. The conducted ROC-analysis has demonstrated high sensitivity and specificity of the mathematical model, which enabled us to recommend determination of IL-6 in the saliva of the children suffering from CT as an additional diagnostic criterion.