The problem of predicting corporate failure has intrigued many in the investment sector, corporate decision makers, business partners and many others, hence the intense research efforts by industry and academia. The majority of former research efforts on this topic focused on manufacturing companies with considerable assets commensurate with their size. But there is a dearth of publications on predicting non-manufacturing firms’ financial difficulties since these firms typically do not have significant assets that rely heavily on assets, and a key variable cannot be adequate. Recently, data envelopment analysis (DEA) rather than Altman’s Z score model and traditional parametric methods has become a research interest in predicting corporate failure. However, there is still no research showing how to fix appropriate cut-off points to distinguish the performance of firms. Our research utilizes slack-based measure (SBM) DEA model to generate efficiency scores for non-manufacturing firms; then we categorize these firms into safe, grey and distress zones by proposing cut-off points based on 5 years DEA analysis. The result shows that the proposed method has obvious advantages in predicting corporate financial stress.