Between 1949 and 1989 the Semipalatinsk nuclear test site (SNTS), an area of 19,000 square km in northeastern Kazakhstan, was the location of over 400 nuclear test explosions with a total explosive energy of 6.6 Mt TNT (trinitrotoluene or trotyl) equivalent. It is estimated that the bulk of the radiation exposure to the population resulted from three tests, conducted in 1949, 1951, and 1953 although estimations of radiation doses received by the local population have varied significantly. Analysis of the published ESR dose reconstruction results for residents of the villages near the SNTS show that they do not correlate well with other methods of dose assessment (e.g. model dose calculation and thermo luminescence dosimetry (TLD) in bricks). The most significant difference in dose estimations was found for the population of Dolon, which was exposed as result of the first Soviet nuclear test in 1949. Published results of ESR measurements in tooth enamel are considerably lower than other dose estimations. Detailed analysis of these results is provided and a possible explanation for this discrepancy and ways to eliminate it are suggested.