The effectiveness of Sample-Entropy was practically shown as the measure index of irregularity from time series, by the development of research to quantify complexity. This Sample-Entropy calculated from at least 100 sample points, can classify a various system which contains a deterministic chaos system and/or a stochastic system, and is strong against noise. Therefore it is applied to the wide fields such as the ecology, finance, the medical treatment, etc. In this paper, the recursive Sample-Entropy technique is proposed and it is applied to the earthquake forecast. The earthquake forecast is generally performed by using the vibration wave, however it is extremely short that time from the forecast to the occurrence. Then, in this research, I use the earth current data based on the technique called the VAN method that used an abnormal earth current that is the one of the macroscopic anomalies that the difference of time until the earthquake occurrence is long. As a result, it is confirmed that there is a possibility that the precursor of an earthquake are able to be observed in real time.