The most established functional MR imaging technique for activation studies relies on a T2*-weighted contrast. This signal arises primarily from a blood oxygen level dependent contrast generated by an imbalance between the increase in regional cerebral blood flow and oxygen metabolism in the brain during activation. As predicted by theory, the percentage signal changes observed in functional MR imaging experiments are considerably smaller than those detected by positron emission tomography, which directly measures regional cerebral blood flow as an index of neuronal activity. Cross-validation of functional MR imaging with an established technique such as positron emission tomography would be extremely valuable for determining the correlation between functional MR image signal change and regional cerebral blood flow change and for assessing the sensitivity of the functional MR imaging technique. The authors report on such cross-validation experiments in three subjects challenged with a verbal working memory task and show that satisfactory replication of positron emission tomography results with functional MR imaging was achieved in two subjects. Limitations owing to magnetic field strength used and single-slice sampling may have contributed to the lack of signal detection in the case where no reliable activation pattern was detected with functional MR imaging.