Background and Purpose-Three-dimensional pseudocontinuous arterial spin labeling with multiple postlabeling delays has been used to assess cerebral blood flow (CBF). We used this modality to estimate antegrade and collateral flow in patients with unilateral middle cerebral artery stenosis. Methods-Consecutive patients with unilateral middle cerebral artery 50% to 99% stenosis at 2 centers underwent pseudocontinuous arterial spin labeling with a postlabeling delays of 1.5 and 2.5 s. Mean CBF of bilateral middle cerebral artery territory at the postlabeling delays 1.5 and 2.5 s was measured. Early-arriving flow proportion was defined as (CBF 1.5 s at lesion side/CBF 2.5 s at normal side)x100%. Late-arriving retrograde flow proportion was defined as ([CBF 2.5 s-CBF 1.5 s] at lesion side-[CBF 2.5 s-CBF 1.5 s] at normal side)/CBF 2.5 s at normal sidex100%. Antegrade and collateral scales were evaluated in patients with conventional angiography. Spearman correlation coefficients were calculated between early-arriving flow and late-arriving retrograde flow proportions on arterial spin labeling and antegrade and collateral scales on conventional angiography, respectively. Results-Forty-one patients (46.0 +/- 12.0 years) were enrolled. The mean early-arriving flow proportion was 78.3 +/- 14.9%. The mean late-arriving retrograde flow proportion was 16.1 +/- 10.2%. In 21 patients with conventional angiography, Spearman correlation coefficient was 0.53 (95% confidence interval, 0.11-0.79) between antegrade grade and early-arriving flow proportion (P=0.01) and 0.81 (95% confidence interval, 0.56-0.92) between collateral grade and late-arriving retrograde flow proportion (P<0.0001). Conclusions-Three-dimensional pseudocontinuous arterial spin labeling with 2 postlabeling delays may provide an empirical approach for estimating antegrade and collateral flow in patients with unilateral middle cerebral artery stenosis.