One of the most important problems related with the multilayer feedforward neural networks is to decide how to choose the most appropriate size of the network for solving a specific problem. That means the number of hidden layers and hidden nodes within each layer since they are significant in characterizing the performance of the network, and it also greatly influences network capacity, generalization ability, learning speed, output error, and output response. If the network size is too small it will not be able to learn the data, on the other hand a large net size may learn very slowly. Pruning algorithm is one of the most popular techniques often used to determine the appropriate net size. The main objective of this work is to present and implement different classical and genetic pruning techniques for solving the problem of determining the appropriate number of nodes in a single hidden layer feedforward network. To accomplish the above work, two different problems (network architectures) are used: a nonlinear function (f(x) = e:" where .Y s[0.1]). and classification net (to classify the input data into three different classes), these two nets must be first trained by BackPropagation training algorithm, and then pinned by different pruning techniques that are categorized into two parts: Classical pruning and Genetic pruning. The classical pruning deals with pruning the net using three different factors: Goodness factor, Consuming energy factor and Weighted power method. The second part deals with pinning the net using different Genetic pruning Algorithms (with different genetic operators: Coding, Selection, Crossover and Mutation). Genetic pinning algorithm is divided into two techniques: weight pruning and Node pruning techniques. After pinning the two nets by classical and genetic pinning algorithms and comparing their results, it has been found that using genetic algorithm to perform pinning process produces results comparable with classical pinning algorithm. When the net size is relatively small then the classical pinning techniques is suitable to pinne the net, but when this size is large the genetic algorithm is an effective way to prune the net.