Natural aging method was used to deal with Baosteel short-flow (BSSF) steel slag, content of free CaO (f-CaO) in BSSF steel slag monitored, and the change mechanism analyzed. Exponential smoothing model based on wavelet analysis was established by wavelet analysis theory and exponential smoothing method to forecast the f-CaO content of steel slag in natural aging. The results showed that f-CaO content of BSSF steel slag was effectively reduced by natural aging method to meet the requirements of safe use, namely f-CaO content in type A steel slag was about 1.2% after 180 d and in type B steel slag about 5.0%~6.0 % after 60 d. Exponential smoothing model based on wavelet analysis predicted data agreed well with the experimental data, the relative error was -3.442%~4.651%, it could increase the prediction accuracy efficiently of the f-CaO content of steel slag in natural aging.