The task of accurately distinguishing between arrivals of different types of seismic waves is a common and important task within the field of seismology. For data generated by seismic stations operated by SNSN this task generally requires manual effort. In this thesis, two automatic classification models which distinguish between two types of body waves, P- and S-waves, are implemented and compared, with the aim of reducing the need for manual input. The algorithms are logistic regression and feed-forward artificial neural network. The applied methods use labelled historical data from seismological events in Sweden to train a set of classifiers, with a unique classifier associated with each seismic station. When evaluated on test data, the logistic regression classifiers achieve a mean accuracy of approximately 96% over all stations compared to approximately 98% for the neural network classifiers. The results suggest that both implemented classifiers represent a good option for automatic body wave classification in Sweden.