The objective of this study was to examine characteristics of non-profitable hospitals, as compared to profitable hospitals, in Thailand. Administration national data for 640 public hospitals of fiscal year 2002 from Ministry of Public Health were analyzed. A two-part model estimated by logistic regression and ordinary least-squares regression was used to examine the likelihood of hospital loss and the amount of loss, respectively. The results showed that various managerial, service mix, and market variables were significantly associated with the likelihood of hospital loss. An increase in average hospital Universal Coverage (UC) inpatient day per number of the UC and the concentration of hospital market could significantly increase the likelihood of hospital loss, while an increase in the