Caffeine is a popular psycho-active drug with effects on the central nervous system and, in particular, on brain haemodynamics. As a non-selective antagonist of the adenosine receptors it increases neuronal firing rates and reduces cerebral blood flow (CBF) via vasoconstriction [1]. Effects on haemodynamics are spatially not homogeneous, as they vary depending on the amount of receptors in each specific area [2]. Previous MRI studies have shown effects on whole grey matter baseline CBF or CBF responses to sensory stimuli in relevant ROIs ([3], [4]), with just one study focusing on the spatial distribution of these effects in resting state ([5]). In this study we examine the changes in CBF and tissue arrival time (TAT) following caffeine consumption, in order to determine in more detail the spatial distribution of these effects. Sixteen healthy, moderate caffeine consumers (between 51 and 298 mg/day; 8 females, age = 24.7±5.1) were recruited for a double-blind, crossover, placebo-controlled study. Each participant was scanned on two separate days (30.1±18.8 days apart, same time of the day). Each scan day (see Fig. 1) included a “pre” capsule multi inversion time (mTI) pulsed arterial spin labelling (PASL) scan followed by the consumption of a 250mg caffeine (or placebo) capsule and two post-capsule mTI PASL session: one about 0.5 h later (“post1”) and another about 1h later (“post2”). Mutiple TI times were used: 0.15, 0.3, 0.45, 0.6 s (short mTI) and 1, 1.4, 1.8, 2.2 s (long mTI). A PICORE tagging scheme was used with a QUIPSS II cutoff at 700ms for long mTIs. A dual-gradient echo (GRE) readout and spiral k-space acquisition imaging was used (TE₁ = 2.7 ms, TE₂ = 29 ms, 64x64, 3x3x7mm³, gap = 1 mm, 12 slices). Salivary caffeine concentration samples were taken to monitor the stability of the post-capsule caffeine levels and the compliance to the request of abstaining from caffeine prior to the scans. Data were pre-processed for motion and physiological noise correction and then fitted with the Chappell two-compartment model [6], resulting in CBF and TAT maps. Estimates of CSF equilibrium magnetization were used for CBF quantification and a low resolution minimal contrast image was used for sensitivity correction. Whole grey matter values of CBF and TAT were calculated for each subject. In order to examine differences between placebo and caffeine conditions voxel-wise t-tests were calculated between the pre and post₂ conditions. Finally changes between pre and post₂ conditions were calculated for both parameters in seven ROIs.Salivary caffeine concentration values underwent an average increase of about 2 mg/l 30 min after caffeine consumption and a further 2 mg/l increase at 80 min, while they were not significantly different from baseline in the placebo condition (Fig. 2). Whole grey matter values of CBF averaged across the subjects significantly decreased by 30.2% (±8%) at post1 and 34.6% (±8.4%) at post₂ with caffeine, compared to non-significant variations of 5.2% (±17.3) and -2.5% (±11.3) respectively with placebo (Fig. 3, top). A significant difference is also found for TAT with an increase of 4.9% (±4%) and 5.1% (±4.9%) compared to 0.9% (±3%) and 0.3% (±3.5) with placebo (Fig. 3, bottom). Neither salivary concentration, nor average daily caffeine intake correlate with CBF or TAT grey matter values (data not shown). T-test maps show a widespread significant decrease in CBF (Fig. 4, top). TAT presents more localized and mixed effects: areas of significant increase are found in medial and inferior brain regions, while areas of significant decrease where smaller and are localized in superior regions (Fig. 4, bottom). The ROI analysis supplies a more complete picture of the distribution of the changes in haemodynamics across the brain (Fig. 5, top), highlighting a seemingly uniform significant response of CBF to caffeine, compared to a more varied variation in TAT (Fig. 5, bottom). This study reports a comprehensive picture of the acute effects of caffeine on brain haemodynamics, at both bulk, regional and voxel-wise levels. Grey matter changes in CBF confirm previous literature reports ([3],[4],[5]), while more extended regional differences are found compared to previous studies ([5]), possibly explained by the more complex ASL model employed here ([6]). TAT results highlight the heterogeneity of the response, whose physiological nature should be investigated further. The regional and voxel-wise variability reported also suggests that great attention must be paid to physiological assumptions when modelling ASL data to estimate CBF in studies on drugs that affect brain haemodynamics.