Ecological correlation analysis was made between the source of drinking water and standardized incidence ratios (SIRs) of malignant neoplasms registered in Niigata, Japan. The purpose of this study was to find out some clues for generating a working hypothesis that the contamination of drinking water with environmental chemicals especially of endocrine disrupting chemicals in big rivers may be related to the occurrence of cancers. Sources of drinking water in the year 1974 were classified into two categories; one from the Shinano and Agano and their down-stream rivers, and the other from small rivers flowing into these big rivers, rivers originating from the mountain areas, underground and reservoirs (The other group). After excluding registered cases under 20 years of age, the registry data (1992-94) were sub-divided into two categories according to the source of drinking water. The SIRs were calculated by using two different standard populations; Niigata total and "The other group". All of the sites were examined in the former standard. In the latter case, however, 12 and 9 sites in males and females, respectively were selected based on the assumption that the expected numbers in each age group totaled 5 or over when there were more than 200 registered patients as a total. Concerning the judgment for the significant association in the former standard, we considered the presence of positive associations when both of the SIRs in the two categories were significantly different from the Niigata total population (p < 0.05 by chi 2 test). Of the 33 sites in males, the following sites were correlated with the source of drinking water; tongue, colon, rectum, gallbladder, etc., prostate, kidney, etc. and all sites. Of 35 sites in females, those were the colon, gallbladder, etc., bronchus and lung, breast, corpus uteri, thyroid and all sites. Of the 12 and 9 respective sites subjected to the second analysis, positive associations were observed in the following: esophagus, colon, rectum, liver, gallbladder, etc., bronchus and lung, prostate, bladder and all sites in males, and colon, rectum, gallbladder, etc., bronchus and lung, breast and all sites in females. Final conclusion, however, should be waived until the consistency of association can be confirmed by using more reliable incidence data, because the difference in the death certificates only (DCO) rate between the two groups might have played as a bias. The use of mortality data may be an alternative way to evaluate the present findings in terms of the consistency of association.