In this article, we consider a stochastic profit model with discounting for a simple delivery scheduling problem of spare units and maximize the expected total discounted profit over an infinite time planning horizon. Then the familiar net present value (NPV) approach is applied to formulate the present value of expected total profit for an infinite time planning span. In a fashion similar to Dohi et al. (2000), we further propose a statistical estimation procedure for the optimal delivery schedule, based on the non-parametric method. More precisely, the concept of total time on test statistics is utilized for the development of a non-parametric estimation method. In a simulation study, we investigate numerically asymptotic properties of the resulting estimator of the optimal delivery schedule. It can be concluded that the developed method here can provide nice convergence properties to the real optimal solution and be useful for actual spare parts management