Kinetic modelling aims to provide more specific information on in vivo physiology than simple image quantification approaches based on a single image, such as standardised uptake value (SUV) and target-to-background ratio (TBR). They can be divided into two classes depending on whether they are based on a compartmental model or not. Compartmental modelling assumes that the kinetics of the PET radiotracer can be well described by transfer between compartments that represent different states of the tracer. As the number of free parameters that can be determined is limited, typically no more than two compartments are used to model the tissue signal. An alternative to using a simplified model, which maybe inaccurate, is to use a model-independent approach, the most widely used of which are graphical techniques (Patlak and Logan plots) and spectral analysis. Model-independent approaches can only directly estimate macro-parameters (K1, VT, DVR and Ki), whereas compartmental modelling can also estimate micro-parameters which maybe more specific (e.g. k3 of FMISO in hypoxia). To obviate invasive blood sampling, versions of model-based and model-independent techniques have been developed which use a reference tissue input rather than a blood-based input function. This presentation will describe the main kinetic modelling approaches, the data that they require, and their relative advantages and disadvantages.