Seasonal rainfall predictability over the Huaihe River basin is evaluated in this paper on the basis of 23-year (1981–2003) retrospective forecasts by 10 climate models from the Asia-Pacific Economic Cooperation (APEC) Climate Center (APCC) multi-model ensemble (MME) prediction system. It is found that the summer rainfall variance in this basin is largely internal, which leads to lower rainfall predictability for most individual climate models. By dividing the 10 models into three categories according to their sea surface temperature (SST) boundary conditions including observed, predicted, and persistent SSTs, the MME deterministic predictive skill of summer rainfall over Huaihe River basin is investigated. It is shown that the MME is effective for increasing the current seasonal forecast skill. Further analysis shows that the MME averaged over predicted SST models has the highest rainfall prediction skill, which is closely related to model’s capability in reproducing the observed dominant modes of the summer rainfall anomalies in Huaihe River basin. This result can be further ascribed to the fact that the predicted SST MME is the most effective model ensemble for capturing the relationship between the summer rainfall anomalies over Huaihe River basin and the SST anomalies (SSTAs) in equatorial oceans.