Arterial spin labeling (ASL), also known as arterial spin tagging, is a magnetic resonance imaging technique used to quantify cerebral blood perfusion by labelling blood water as it flows throughout the brain. ASL specifically refers to magnetic labeling of arterial blood below or in the imaging slab, without the need of gadolinium contrast. A number of ASL schemes are possible, the simplest being flow alternating inversion recovery (FAIR) which requires two acquisitions of identical parameters with the exception of the out-of-slice saturation; the difference in the two images is theoretically only from inflowing spins, and may be considered a 'perfusion map'. The ASL technique was developed by John S. Leigh Jr, John A. Detre, Donald S. Williams, and Alan P. Koretsky in 1992.

Arterial spin labeling utilizes the water molecules circulating with the brain, and using a radiofrequency pulse, tracks the blood water as it circulates throughout the brain. After a period of time in microseconds (enough to allow the blood to circulate through the brain), a 'label' image is captured. A 'control' image is also acquired before the labeling of the blood water. A subtraction technique gives a measurement of perfusion. In order to increase SNR, collections of control and label images can be averaged. There are also other specifications in the MRI that can increase SNR, like the amount of head coils of the MRI, or a stronger field strength (3 T is standard, but 1.5 T is satisfactory). In order to properly scale the perfusion values into cerebral blood flow units (CBF, ml/100g/1 min), a separate proton density map with the same parameters (but longer TR to fully relax the blood spins) is recommended to be acquired as well. Alternatively, the average control image can be used to generate CBF, which is the case for Phillips pCASL readouts. Usually background suppression is also applied to increase the SNR. Due to the different variations of each implementations, it is recommended that a large multi-scanner study should design a protocol minimizing the variety of readout methods used by each scanner.

One study has shown that although there are voxel differences when different readout methods are used, average gray matter CBF are still comparable. Differences in SNR are apparent when each voxel compared, but collectively are negligible.^{[importance of example(s)?]}

In continuous arterial spin labeling (CASL), the blood water is inverted as it flows through the brain in one plane. CASL is characterized by one single long pulse (around 1–3) seconds. This may be disadvantageous for certain scanners that are not designed to maintain a radiofrequency pulse that long, and therefore would require adjustments to a RF amplifier. This is rectified in pseudo-continuous arterial spin labeling (pCASL), where a single long pulse is replaced with multiple (up to a thousand) millisecond pulses. This leads to a higher labelling efficiency. pCASL is the preferred implementation of ASL. There are different readout modules for pCASL, depending on the scanner used, with 2D pCASL usually being implemented for all scanners and 3D pCASL stack of spirals implemented in GE scanners.

In pulse arterial spin labeling (PASL), blood water is inverted as it passes through a labeling slab (of 15 to 20 cm) instead of a plane. There are different variations of this implementations, including EPISTAR and PICORE and PULSAR. Most scanners have been designed to have PASL work out-of-the-box for research use.

Velocity selective arterial spin labeling is a strategy that still requires validation. Velocity selective arterial spin labeling is advantageous in a population where blood flow may be impeded (e.g. stroke), because the labeling occurs closer to the capillaries. This allows the post labeling decay to be shorter.

Diffusion-prepared pseudocontinuous ASL (DP-pCASL) is a more recent ASL variant sequence that magnetically labels water molecules and measures their movement across the blood-brain barrier complex, which allows for the calculation of the water exchange rate (kw). kw is used as a surrogate for BBB function and permeability. Water exchange across the BBB is mediated by a number of processes, including passive diffusion, active co-transport through the endothelial membrane, and predominantly by facilitated diffusion through the dedicated water channel aquaporin-4 (AQP4). Several studies have investigated the use of DP-pCASL in cerebrovascular diseases, including acute ischemic stroke, CADASIL, hereditary cerebral small vessel disease as well as in animal models.

ASL maps can mainly be analyzed using the same tools to analyze fMRI and VBM. Many ASL-specific toolboxes have been developed to assist in ASL analysis, such as BASIL (Bayesian inference for arterial spin labelling MRI), part of the FSL neuroimaging package and also Ze Wang's ASL toolbox (using MATLAB) to assist in the subtraction and averaging of the tagged/control pairs. A visual quality check is often needed to make sure that the perfusion map is valid (such as correct registration, or correct segmentation of non-cerebral materials such as the dura mater). A whole brain/voxel-wise approach can be analyzed by registering the ASL map into MNI space for group comparisons. A region of interest approach can be analyzed by registering the ASL map into a selected cluster, or an atlas, like a standard (such as the Harvard-Oxford Cortical atlas) or an individual atlas developed by software like FreeSurfer. The recommended procedure of ASL registration for voxel-wise analysis is to register the perfusion map to a gray matter segmentation of each individual in a non-rigid procedure.^{[citation needed]}