Obtaining satellite clock bias (SCB) in real-time is a key problem in real-time precise point positioning (RTPPP). A repair method specific to SCB correction derived from the International GNSS Service (IGS) real-time service (RTS) is proposed, and the stability and precision of GPS SCB correction in the real-time data stream of IGS02 and IGS03 are analyzed. Based on the characteristics of SCB correction, a new SCB correction prediction algorithm is theoretically derived and established by making a single difference for the provided data. The new prediction algorithm is compared with the linear model, quadratic polynomial model, and grey model in prediction tests. The results show that the prediction accuracy of the new algorithm is the best, with a prediction accuracy of 0.06 ns and prediction length of 30 s, meeting the requirements for RTPPP.