Data Envelopment Analysis is widely used in the field of hospital efficiency measurement. The result can be influenced by the choice of input variables. Two groups of input variables are applied in the analysis. Group 1 is consisted of financial variables: labor cost, non-labor cost and capital cost. Group 2 are non-financial variables including number of doctors, nurses, other personnel and beds. Input-orientated Data Envelopment Analysis is used to measurement the technical and scale efficiency of three level hospitals (regional, general and community hospital) in Thailand of 2010 with two different groups of input and the same output. The result reveals that for regional hospital, the choice of input variables has no influence on the measurement of efficiency. For general hospital, the choice of input has no influence on the measurement of technical efficiency, but has special impact on the measurement of scale efficiency. The scale efficiency scores come from two models with different inputs of hospitals are different from each other. This reveals that general hospitals have had enough money but not enough labor and bed. They need to increase their scale by hiring more employees and opening more beds. For community hospital, the result of DEA is not steady and the scores change when different input variables is chosen.