An optimal accelerated life test plans is developed under periodic inspection and Type I censoring when the lifetime is exponentially distributed. The mean lifetime is assumed to be an exponential function of a stress, and the maximum likelihood estimation is used to obtain the estimates of unknown parameters at the use stress. The criterion adopted is the asymptotic variance of the estimated log mean lifetime at the use stress, and the decision variables consist of the low stress level and portions of test items allocated to each stress. Computational experiments are conducted to observe how optimal plans vary with respect to failures probabilities at the high and use stress and the number of inspections.