In order to forecast the small sample large oscillating sequences, this paper proposes a grey interval forecasting method. Firstly, two non-equidistance GM(1,1) models are built for upper and lower sequences respectively, and the development boundary of the system are described by the upper and lower envelope curves. Then, the computing method for the interval and basic forecasting values of the original sequence are proposed, and the algorithm is constructed. Finally, the grey exponent law and timeliness of interval forecasting model are studied. The numerical experiment shows that the value of the development coefficient is not the only factor influencing the timeliness. The application example shows that the forecasting accuracy can be effectively enhanced.