By taking the random parameters into consideration, a prediction approach for the service life of subsea tunnels is proposed based on the reliability theory. The proposed method generated the failure model for the lining system according to the failure of outside or inside of a lining. The failure model takes the corrosion of reinforcing bars as a signal of losing the durability, in which the chloride corrosion is considered for the outside of the lining and both the chloride corrosion and carbonation are considered for the inside of the lining. The time-varying reliability index for the prediction of service life is calculated by Monte Carlo method. The results show that the chloride corrosion has a severer influence than the carbonation on the service life of tunnel lining, and it’s more realistic to estimate the service life of subsea tunnels by the prediction method based on the system reliability index.