The spontaneous ignition behavior is considered as one of the most important factors to the success of air injection process (AIP) application. Although the theory of crude oil spontaneous ignition is easy to understand, this phenomenon turned out to be difficult to capture in the lab. In this study, the Frank-Kamenetskii method was applied to investigate the spontaneous ignition of the crude oil sample both theoretically and experimentally. The thermal experiments were performed to study the thermal oxidative characteristics of crude oil samples and obtain their kinetic data. The kinetic data and typical reservoir properties were used to predict the spontaneous ignition, and the oven tests with a mixture of crude oil and sand samples were performed to validate the predictions. The experimental results of crude oil spontaneous ignition failed to prove the predictions made by the Frank-Kamenetskii theory, which implies that the complicated oxidation reactions of crude oil cannot be considered as one single-step reaction with a constant fuel ratio in the low-temperature oxidation stage, and the uniformly distributed heat within the body of the sample also needs to be reconsidered. Also, this study shows that the coke does not produce tremendous heat under low-temperature conditions, although the coke may be important for sustaining the combustion after the combustion front was already developed, and it does not contribute to the spontaneous ignition process. Moreover, this study shows that the reason for spontaneous ignition failure may be because the gaseous light oil components escaped during tests, which is different compared to the real AIP practice in reservoir conditions, and the gaseous light oil components are the key for ignition. Therefore, a conventional reaction model for in situ combustion needs to consider the gaseous phase combustion, especially when predicting the spontaneous ignition.