Early diagnosis of hepatocellular carcinoma (HCC) is still a great challenge in clinical practice. Tumor markers such as alpha-fetoprotein(AFP), liver enzymes, cytokines, and some special glycoproteins, though helpful, are not sensitive and specific enough for early diagnosis of HCC. The establishment of several interesting predictive diagnostic models on liver fibrosis/cirrhosis suggests that mathematic predictive model, which is developed based on large sample size and follow-up study, might be of higher sensitivity, specificity and feasibility in clinical application. Here we suggest that more attention should be paid to this kind of multi-parameter predictive diagnostic models clinically, so as to improve the early diagnosis of HCC in a more economical and feasible way.