Проведен межлабораторный термогравиметрический эксперимент по идентификации огнезащитных составов интумесцентного типа. Использование дисперсионного анализа позволило установить значимость/незначимость различий в поведении исследуемых материалов при нагревании, выявить лаборатории, обладающие необходимыми компетенциями для проведения идентификации. Найдено, что применение различных водно-полимерных дисперсий даже близкого строения в составе огнезащитных материалов приводит к неидентичности их свойств. Представленный подход позволил однозначно установить неидентичность исследованных огнезащитных составов. The article deals with the problem of identification of materials, substances and means of fire protection according to GOST R 53293-09 using one of the methods of thermal analysis, namely, thermogravimetric analysis. The long-term practice demonstrated that the methodology presented in the standard did not always give a clear result when identifying objects. Therefore, the evaluation experiment was conducted simultaneously in several testing laboratories with the appropriate equipment. The tests were conducted in the laboratories of the FGBU VNIIPO EMERCOM of Russia, ANO «CISIS FMT», and two CCPS of the Lomonosov Moscow State University. The most complex and controversial objects for identification were studied, namely, thermally expanding (intumescent) flame retardants based on water-polymer dispersions, which are used to protect building structures in closed premises. The purpose of the work consists in more correctly identifying the differences in the samples of the identifier and the analyzed object by using methods of dispersion analysis (GOST R ISO 5725-1, 2-202) when processing the joint interlaboratory experiment. Thermogravimetric analysis of objects was performed in platinum crucibles on samples in the form of disks with a thickness of 0,2-0,3 mm, weight 5-7 mg, at a heating rate of 20 °C/min in the air flow (50-70 ml/min). As a result of the experiment, the significance/insignificance of differences between samples should be displayed. The opinion about the competence of the laboratories participating in the experiment was made. In each laboratory there were conducted two series of thermogravimetric tests in 30-700 °C temperature range including five parallel measurements in each series. Each independent single measurement for further processing was represented by the data array formed within the temperature range of 140-650 °C with a 5° step, in total 103 points in each measurement. The results were processed using Excel spreadsheets. In the context of dispersion analysis there were evaluated both mean values of conversion rates for all laboratories, and the convergence (repeatability) variance for each laboratory using the Student's test, and the inter-laboratory variance (reproducibility variance) using the Cochran’s С test. This procedure revealed the nature of the quasi-samples and established the identity/non-identity of the samples. The results of calculations are represented by the graphical dependence of the generalized average degrees of transformation of destruction processes of objects under the temperature as well as the difference between these characteristics for laboratories I-IV. The presented identification procedure allows both to uniquely solve the problem of significance/insignificance of differences between the test object and the identifier, and to distinguish samples that have a similar composition. The statistical characteristics of a single measurement are calculated. The convergence of parallel measurements for determination of statistical outliers is identified. Determination of the overall average difference for a series of degrees of transformation allows to calculate the Student's test and accept or reject the null hypothesis. The performance of the joint evaluative (interlaboratory) experiment allows not only to establish the significance/insignificance of differences in the behavior of the materials during heating, but also to identify laboratories that have the necessary competencies for identification. The case considered in this paper revealed that the use of various water-polymer dispersions of even close structure in the composition of fire-resistant materials leads to non-identity of their properties.