OBJECTIVES: Observational study is difficult to interpret due to a number of methodological issues; cohort effects, the learning curve of the health care provider and confounding by indication: This large scale retrospective study is aimed to ascertain that there are indication bias of LASIK surgery between local hospital and tertiary hospital in South Korea. METHODS: A retrospective cohort for consecutive 3,401 eyes received LASIK in 6 multicenter including tertiary hospital and local hospital from 2002 to 2005 were registered. Sociodemographic, operation characteristics, preoperative baseline manifest refractions(MR) and additional follow up data at 3month, 6month and 12 month after surgery were collected. We used independent t-test to compare preoperative difference between tertiary hospital and local hospital and generalized linear mixed model with random intercept to test characteristic of hospital and follow up time might influence to MR after surgery. RESULTS: The baseline MR (mean±SD) of tertiary hospital was -5.24±2.07 and that of local hospital was -4.27±1.71 (p<0.001). The mean MR at 3 month, 6 month and 12 month of tertiary hospital were -0.46±0.70, -0.55±0.69 and -0.63±0.67 increasingly according to the follow up duration. The mean MR at 3 month, 6 month and 12 month of local hospital were -0.48±0.62, -0.43±0.52 and -0.47±0.47. A generalized linear mixed model with a random intercept for MR change revealed that the higher manifest refraction was significantly influenced by the longer follow up time(p<0.001) but the hospital characteristic was not influence to MR(p=0.319). No significant interaction between follow up duration and level of medical institution was detected. CONCLUSIONS: Although observational study might be confounded by indication bias, no evidence of such bias was detected. It may reflect real world and assessment of indication bias still provide us with benefits on clinically relevant interpretation about current experience of medical interventions.