A correlation is established between the depths of incomplete-fusion areas along the surfaces of welded flanges in T-joints and incomplete structural fusion. A rejection level is chosen for the echo-signal amplitude that makes it possible to decrease errors in ultrasonic testing.In accordance with Decree no. 241 of the Russian Government as of March 28, 2001, systems for nondestructive testing of technological facilities and structures that are used and operated at hazardous production establishments have been developed and implemented [1] as part of the Expert Analysis System for Industrial Safety. During the manufacture, construction, repair, installation, and technological testing, nondestructive testing methods are most often used for checking welded joints. The widely-used ultrasonic testing of welded joints may be complicated by incomplete structural fusion, that is, incomplete fusion that occurs due to the welded-joint design itself [2,3]. T-joints belonging to type T1 (GOST 5264-80, GOST 8713-79, and GOST 14771-76) are the weld joints where structural incomplete fusion may occur and that are most widely applied in the design of cargo cranes, lifting devices, rigs, and mining machinery.Cuts made in T1-type welded T-joints demonstrated that incomplete structural fusion located in weld roots along the surfaces of welded flanges is the most typical internal flaw in these elements. It should be noted that echo signals are generated by only one component of incomplete fusion, namely, incomplete fusion at the surface of a welded element. There is, however, a rather strong correlation between the dimensions of the incomplete-fusion areas located at different surfaces of the welded flanges (Fig. 1).If there is an incomplete-fusion area at one surface, an incomplete-fusion area is usually observed at the other surface as well. Moreover, the dimensions of incomplete-fusion areas are close to each other. To analyze the incomplete-fusion topography and to determine the correlation dependence between the dimensions of an incomplete-fusion area and the echo-signal amplitude, we used specimens in which incomplete fusion was provoked, in addition to specimens cut from actual objects and welded using standard technology. The welding speed was artificially increased in the latter case and the welding torch was operated in a special way.Reliability is the most important indicator of testing efficiency. Testing reliability R is the probability of making a correct decision when assessing the product quality [4]. In order to determine reliability indicators, it is necessary to assess the possible testing errors. The first-type of error involves supplier risk α (over-rejection), and the second-type of error involves consumer risk β (under-rejection). The testing reliability can be calculated using the following formulas:Reliability indicator R α takes into account over-rejection, and R β takes into account under-rejection. Indicator R Σ is the most widely used and takes into account both over-and