In positron emission tomography studies using bolus injection of [15O]water, activation responses reflect underlying CBF changes during a short time (15 to 20 seconds) after arrival of the bolus in the brain. This CBF sensitivity window may be too short for complex activation paradigms, however, particularly those of longer duration. To perform such paradigms, we used a slow infusion method of tracer administration to lengthen the CBF sensitivity window. The present study was designed to determine if this slow infusion technique yields similar results to a bolus injection with a short activation task involving memory for faces. When analyzed using statistical parametric mapping, scanning durations of either 90 or 120 seconds and a 90-second slow infusion schedule produced very similar results to a standard 60-second scan collected after bolus injection, indicating that statistically similar brain activation maps can be produced with the two infusion techniques. This slow infusion approach allows for increased flexibility in designing future studies in which a short CBF sensitivity window is a limiting factor.