This document describes oblique trees classification, a new algorithm for tree-based classification that uses oblique or nonlinear splits of the data rather than restricting splits to individual variables. It was found to outperform classical tree algorithms using only single variable splits. The oblique tree algorithm works by finding the best splitting planes in multi-dimensional space using a Gini-based criteria. Simulations on both synthetic and real-world datasets demonstrated that oblique trees consistently had a smaller error rate and required fewer trees to achieve a desired accuracy level of classification compared to classical orthogonal trees. Future work may include handling high-dimensional data and providing software implementation for users.