Objective To explore the establishing methods and differences of rat models of spleen deficiency and spleen deficiency liver cancer using the traditional Chinese medicine Dachengqi and Xiaochengqi decoctions.Methods Spleen-deficiency rat models were developed by multifactor methods: bitter-cold purgation (Dachengqi or Xiaochengqi decoction), cold-wet environment, tiredness, and fasting on alternate days for 30 days. Seven days after spleen-deficiency modeled,liver cancer in the spleen-deficiency rats and normal rats was developed by subcutaneously inoculation of Walker-256 carcinoma cell line in nude mice and then transplanted into rat livers. Liver cancer models were observed for 35 days. Sixty 3-week old male Wistar rats were randomly distributed into 4 groups: normal group, liver cancer model group, and Dachengqi and Xiaochengqi decoction groups. Degree of spleen deficiency, changes of the body-weight, survival time and tumor formation were recorded. Results Spleen deficiency rat models were successfully established. The weight gain of rats in the spleen-deficiency groups was significantly inhibited (P<0.01), and during the first 20 days (but not later) the average body weight of the Dachengqi decoction group was significantly higher than that of the Xiaochengqi decoction group (P<0.05). Spleen-deficiency scores of rats in the Xiaochengqi and Dachengqi groups were higher than those in the blank tumor group, especially in the Xiaochengqi group (P<0.01). The total tumor formation rate was 91.1% and 80% in the blank tumor groups, and 93.3% in both Xiaochengqi and Dachengqi groups, respectively. The average survival time of Xiaochengqi group was lower than that of the blank tumor and Dachengqi groups (P<0.01 and P<0.05). The cumulative survival rate of the Xiaochengqi group and rats with a higher spleen-deficiency score was lower than that of the other groups (P<0.05). Conclusions Xiaochengqi decoction may induce spleen deficiency more seriously than Dachengqi decoction, and spleen deficiency may be an important unfavorable prognostic factor for rat models of liver cancer.