We describe here experimental considerations in the implementation of quantitative perfusion imaging techniques for functional MRI using pulsed arterial spin labeling. Three tagging techniques: EPISTAR, PICORE, and FAIR are found to give very similar perfusion results despite large differences in static tissue contrast. Two major sources of systematic error in the perfusion measurement are identified: the transit delay from the tagging region to the imaging slice; and the inclusion of intravascular tagged signal. A modified technique called QUIPSS II is described that decreases sensitivity to these effects by explicitly controlling the time width of the tag bolus and imaging after the bolus is entirely deposited into the slice. With appropriate saturation pulses the pulse sequence can be arranged so as to allow for simultaneous collection of perfusion and BOLD data that can be cleanly separated. Such perfusion and BOLD signals reveal differences in spatial location and dynamics that may be useful both for functional brain mapping and for study of the BOLD contrast mechanism. The implementation of multislice perfusion imaging introduces additional complications, primarily in the elimination of signal from static tissue. In pulsed ASL, this appears to be related to the slice profile of the inversion tag pulse in the presence of relaxation, rather than magnetization transfer effects as in continuous arterial spin labeling, and can be alleviated with careful adjustment of inversion pulse parameters.