The development of detailed chemical kinetic models has proceeded unabated since the pioneering work of Dixon-Lewis and coworkers forty-five years ago. In that time, computational power has increased ten million-fold, and yet, we do not have a consensus on a complete description for the pyrolysis and oxidation of fuels as simple as hydrogen. This paper presents a comparison of three detailed models and one lumped, semi-empirical model, all from literature, for the oxidation of normal heptane, a fuel of practical interest because of its use in gasoline surrogates. Specifically, the ignition delay times and species concentration histories predicted by the models are compared. We find that most of the models give predictions that agree with each other within the propagated uncertainty from the rate constant estimates, and also show "satisfactory agreement" with experimental measurements. Differences among the models can be attributed to peculiarities in how the models treat the initial decomposition of the fuel. The rate coefficients for these reactions do not agree within the uncertainty assigned for them, however, which means that the models are fundamentally, quantitatively different. The model's predictions agree with each other and with experiment, however. The predictive ability of the models is the only way we can fairly compare them, and by that measure all of the models are equally good. Given the predictive capability of the semi-empirical model, we suggest that future combustion research should concentrate on developing a model that treats only the key, rate-limiting processes in a rigorous manner, rather than continuing the current trend of attempting to describe all possible elementary reactions exactly. Published by Elsevier Inc. on behalf of The Combustion Institute.