Small area estimation is becoming increasingly important because of recent ,growing demand for reliable small area statistics. In Taiwan, the current estimations of labor force series for local governments have unacceptably large ,standard errors due to unduly small size of the sample in the sub-areas. Mixed,effects models have been proposed to improve estimates for given small areas by ,borrowing strength from neighboring areas. In .this paper, we proposed mixed ,effects model to solve the small area estimation problem by combining cross-sectional and time series data. We also classify small areas into several groups ,by their characteristics to overcome the lack of auxiliary data. Our mixed effects ,model can reduce variation of the current survey estimates and produce more ,reasonable results.