The autooxidation of hydrocarbons and oxygenates below ~900 K is controlled by a set of chain reactions that depend strongly on the nature of the parent fuel molecule. In this low-temperature regime, bulk properties such as the autoignition time depend on molecular structure in a complex way. Moreover, many of the elementary reactions happen on multiwell potential energy surfaces that causes these reactions to depend not just on temperature, but on pressure as well. After explaining the chemical steps that control autooxidation under these conditions and the theoretical methods that we use to describe them, I will focus on recent aspects that I studied, often in close collaboration with experimentalists at Sandia. I will show the example of water elimination in case of radicals derived from alcohols to highlight the need and difficulties when it comes to discovering new pathways. Exploring multidimensional potential energy surfaces in an efficient way is a hot area of research and requires a combination of strategies. I will describe the computational tool called KinBot that I am developing to automatically discover and systematically characterize chemical pathways for a wide variety of reactions. I will also discuss our work that led to the detection of an ephemeral, but key radical, called QOOH, and talk about the theoretical challenges in these types of studies.