The starting point of a transport development project is knowledge on travel demand. If there is time and money for doing a detailed survey and passenger counting, we are able to build up detailed travel demand models. Usually this is not the case of the reality. Normally there is money and resources for some counting and maybe some surveying to clarify the demand data (e.g. OD matrix). There is always a discussion point; what is the needed size of the sample at a surveying, to be able to produce reliable data for OD matrix estimation. The paper deals with the issue of needed sample size and reliability of results.