Conventional data analysis concentrates on individual surveys of not more than 2000 respondents. Hypotheses regarding small sub-groups of the population cannot be tested with these surveys. Special surveys with disproportionate samples or with a sample size of over 2.000 respondents are expensive. The cumulation of surveys for the use of secondary analyses can be regarded as a first step towards the solution of this problem. The following problems of the cumulation of surveys are discussed in this paper: 1. Technical problems concerning conditions which single surveys have to fulfil for cumulation, and questions of computer equipment to handle a large amount of data; 2. cost problems; 3. sampling problems, e.a. estimates of single surveys in comparison with estimates of a cumulated survey; 4. problems of equivalence of indicators. Finally other possibilities of analyses with cumulated surveys are discussed.