Short-term regulation of cerebral blood flow (CBF) is controlled by myogenic, metabolic and neurogenic mechanisms, which maintain flow within narrow limits, despite large changes in arterial blood pressure (ABP). Static cerebral autoregulation (CA) represents the steady-state relationship between CBF and ABP, characterized by a plateau of nearly constant CBF for ABP changes in the interval 60–150 mmHg. The transient response of the CBF–ABP relationship is usually referred to as dynamic CA and can be observed during spontaneous fluctuations in ABP or from sudden changes in ABP induced by thigh cuff deflation, changes in posture and other manoeuvres. Modelling the dynamic ABP–CBFV relationship is an essential step to gain better insight into the physiology of CA and to obtain clinically relevant information from model parameters. This paper reviews the literature on the application of CA models to different clinical conditions. Although mathematical models have been proposed and should be pursued, most studies have adopted linear input–output (‘black-box’) models, despite the inherently non-linear nature of CA. The most common of these have been transfer function analysis (TFA) and a second-order differential equation model, which have been the main focus of the review. An index of CA (ARI), and frequency-domain parameters derived from TFA, have been shown to be sensitive to pathophysiological changes in patients with carotid artery disease, stroke, severe head injury, subarachnoid haemorrhage and other conditions. Non-linear dynamic models have also been proposed, but more work is required to establish their superiority and applicability in the clinical environment. Of particular importance is the development of multivariate models that can cope with time-varying parameters, and protocols to validate the reproducibility and ranges of normality of dynamic CA parameters extracted from these models.