[en] Full text: Epidemiological follow up studies require the following basic data: exposure, people-years of observation, and registered events (incidence of diseases and/or deaths). Besides, it is important to register possible confounding factors such as sex, age, ethnicity, occupation, etc. Performing a joint Kazakhstan-Japanese project we are collecting such kind of data based on various archive documents. Retrospective radiation exposure reconstruction uses estimation methods with input data obtained from historic data, at present physical measurements and at present bio dosimetric measurements. Epidemiological studies is an important part of dose reconstruction, they provide the data on address history and occupation (for example, out-door or in-door job) in the period of atmospheric nuclear tests in 1949-1963. These data can be obtained from Household Books, which have been created and used by authorities for taxation purposes. Each Household Book cover a 2-3 year time band and contains lists of households (usually families), names, birth dates, occupation, ethnicity, and property. Unfortunately, Household Books have numerous mistakes in name spelling, birth dates, and handwriting is often bad. In addition, there are many cases when small children and old people are not mentioned in the lists. Correct data can be reconstructed by comparing the data of several subsequent Household Books and the data of ZAGSes (civilian registry offices). Such ZAGS document as birth certificate contains not only data on a newborn, but also correct names and birth dates of parents, place of birth. Marriage certificates include correct data on names and birth dates. If still there are any doubts, it is possible to collect additional information by interviewing old resident. Usually reconstruction of these data on a single person requires referral to 19 archive documents. A computer software designer A.Gaskov together with epidemiologists have developed a chain of technological operations and software for collection, reconstruction and input of epidemiological data. Numerous incorrect name spells lead to accumulation of double records in the electronic database. The necessity arose in development of a special technology and software for double record search. Based on the above described technology we have already collected data on more than thirty thousand 1949-1963 residents of Dolon, Znamenka, Kainar and Karaul rural areas. A big job has been already done connected with their subsequent follow up till now. Three possible outcomes are registered: (1) alive till the end of study, (2) lost for follow up due to death (death dates and causes are registered based on death certificates), (3) lost for further follow up due to change of residence place (the date is obtained from a Household Book). Old residents interview allow for obtaining some additional data. The question exists of the reliability of death causes registered in death certificates. This was a subject of an investigation made together with experts from Tokyo Red Cross Hospital Drs. Oritu and Fijiwara. Oncological death causes registered in death certificates were compared with available archive materials of the Oncology Center. The difference was found negligible. Thus, there are archive documents and a technology has been developed, which allow creation of a retrospective cohort and follow up study. A large epidemiological database have been already collected