The injection experiment conducted at the Rangely oil field, Colorado, was a pioneering study that showed qualitatively the correlation between reservoir pressure increases and earthquake occurrence. Here, we revisit this field experiment using a mechanistic approach to investigate why and how the earthquakes occurred. Using data collected from decades of field operations, we build a geological model for the Rangely oil field, perform reservoir simulation to history match pore‐pressure variations during the experiment, and perform geomechanical simulations to obtain stresses at the main fault, where the earthquakes were sourced. As a viable model, we hypothesize that pressure diffusion occurred through a system of highly permeable fractures, adjacent to the main fault in the field, connecting the injection wells to the area outside of the injection interval where intense seismic activity occurred. We also find that the main fault in the field is characterized by a friction coefficient —a value that is in good agreement with the classical laboratory estimates conducted by Byerlee for a variety of rock types. Finally, our modeling results suggest that earthquakes outside of the injection interval were released tectonic stresses and thus should be classified as triggered, whereas earthquakes inside the injection interval were driven mostly by anthropogenic pore‐pressure changes and thus should be classified as induced.