In this work, a new pattern classification model, working in supervised mode, is presented. The design and operation of this model is based on the Heaviside function. Also, this Heaviside Classifier is of the one-shot kind, which guarantees that the new model will not have any convergence problem. In order to achieve the former, two original operations, called L and C, are proposed; the Heaviside function strongly intervenes in the design of these operations. The pattern learning phase of the new model is based on the original operation L, while the pattern classification phase relies on the effectiveness of the new operation C. With the goal of theoretically substantiating the Heaviside Classifier, some lemmas, theorems, and corollaries are stated and proved. These theorems exhibit relevant properties of the new operations, which in turn affect directly the performance of the new model. In preliminary experimental tests, included in this paper, the Heaviside Classifier has been applied to some data sets known and used by the international academic community. The data obtained from the experimental tests show that the new model performance is competitive, and in some cases superior, with respect to outstanding models in the state of the art on topics related to Computer Intelligence, Data Mining, Pattern Recognition, and Pattern Classification (in the supervised learning mode)