Artificial neural networks (ANNs) originated in the 1940s, when McCulloch and Pitts tried to create a mathematical model of the biological neuron. In 1958 Frank Rosenblatt proposed the development of the multi-layer perceptron, ushering in an era of optimism and intensified work in this area. In short, ANNs receive input signals and multiply them by the weights of the connections, add up the products of the previous multiplication, perform the activation, generating the binary output that represents the activation of this neuron. Unlike ordinary computer programs, they allow learning from observational data and fulfillment of tasks. They have the advantage of dealing with complex, non-linear data and the ability to learn hierarchical representations of the data. ANNs have been used extensively in the pharmaceutical sciences, especially in pharmacokinetics (PK) and pharmacodynamics (PD). The performance of ANNs modeling was superior to linear modeling for PK/PD prediction.