Adaptive autoreclosing is a fast emerging technology for improving power system marginal stability during faults. It avoids reclosing unto permanent faults and recloses unto transient faults only after the secondary arc has extinguished. The challenges that come with the application of the adaptive autoreclosing technology are enormous. To come to grips with these challenges, researchers have been proposing autoreclosure schemes which use artificial neural network (ANN), the reason being that ANNs have in recent years clearly demonstrated their ability in solving some long standing problems in power systems where conventional techniques have difficulty. This paper proposes one such scheme for single-pole autoreclosure. The scheme uses multilayer perceptron artificial neural network which decides whether a fault is transient or permanent based on the percentage of energy in detailed coefficients of Daubechies db4 mother wavelet of the power system faulted voltage. In the case of transient faults, the neural network further determines the optimal reclosure time. A multilayer perceptron neural network was developed to suit the input signal adopted for the scheme and trained using the Levenberg-Marquardt back-propagation technique. The scheme was simulated using the Electromagnetics Transient Programme (EMTP) and MATHLAB software. The results of the simulation show that the proposed ANN-based adaptive single-pole autoreclosure (AdSPAR) scheme is capable of distinguishing between permanent and transient faults and in the case of the latter, predict optimal reclosure times.