A universal, multifunctional system of environmental diagnostics of heat engines and boiler plants has been created, which allows one to determine indicators characterizing the chemical and physical effect of these objects on the environment: concentrations, mass, specific and average operational emissions of pollutants, noise, thermal pollution, vibration. This measuring system consists of instrumental, testing, demonstration and laboratory modules, which allows you to use it as a diagnostic tool, training and test bench and laboratory; as well as apply it in various fields: transport, energy, environmental and educational fields. The diagnostic system implements methods for monitoring and improving the accuracy of measurements of average operating emissions of pollutants: method for determining the resulting measurement errors of the average operational emissions of gaseous pollutants and particulate matter – GAS and РT indicators, which allows you to evaluate the impact on the data of the value of the errors of the measuring equipment of the diagnostic system; a method for increasing the accuracy of measurements of the normalized РT index by taking into account the methodological error of measurements of a given value due to the influence of the temperature of the sample in the tunnel on the measured emission of particulate matter - δРТt. Experimental studies of the diagnostic system and methods for increasing its accuracyon full-scale objects were carried out: diesel engines: tractor 4CHN12/14 tractor D65M, diesel locomotive diesel engine DEL-01 and boiler units: gas - DKVR-20/13 and AOGV-100E, solid fuel - KCHM-2M-4. Transport diesels were tested according to the cycles established by the UNECE Regulations R-49, R-96 and the international standard ISO-8178. As a result of tests of these engines, the coefficients KРi, KMgasi and KMрmi were determined, which are used to assess the accuracy of measurements of GAS and РT indicators, the resulting measurement errors of these values were investigated and the range of variation of the methodical error δРТt was determined.