The numerical simulation of practical combustion devices such as engines and gas turbines requires the coupling of descriptions of complex physical flows with complex chemistry in order to accurately predict phenomena such as ignition and flame propagation. For three-dimensional simulations, this becomes computationally challenging where interactions between large numbers of chemical species are involved. Historically therefore, such simulations used highly simplified descriptions of chemistry, which limited the applicability of the models. More recently, however, a range of techniques for reducing the size of chemical schemes have been developed, where the resulting reduced schemes can be shown to have accuracies which are almost as good as much larger comprehensive mechanisms. Such techniques will be described in this chapter. Skeletal reduction techniques are first introduced which aim to identify redundant species and reactions within a mechanism over wide ranges of conditions. Approaches based on sensitivity analysis, optimization and direct relation graphs are introduced. Lumping techniques are then discussed which exploit similarities between the structure and reactivity of species in describing lumped components, which can represent the sum of several isomers of a particular hydrocarbon species for example. Both approaches can lead to a substantial reduction in the size of chemical mechanisms (numbers of species and reactions) without having a significant impact on model accuracy. They are combined in the chemistry-guided reduction approach, which is shown to generate reduced chemical schemes which are small enough be used within simulations of ignition behaviour in a homogeneous charge compression ignition (HCCI) engine.