Most of the National Statistical Offices, in order to economize the available resources in conducting surveys, optimize the sample size at the national level. However, a need to more detailed information in the level of smaller communities such as provinces, towns, regions, etc is always felt. In a sense, the size of sample should be determined in a way that optimizes the estimators both in the national level and in small area levels. Since rotation sampling is regarded as one of the most important methods of survey sampling in national statistical offices, in this paper, an optimal sample size for a Labour Force Survey -which is performed in each reason by means of rotation sampling by Statistical Center of Iran- is obtained initially, then using sample size determination methods for estimations in small areas, the allocated sample sizes for each sub populations (provinces) are obtained.