This was a blinded crossover study to determine the in vivo distribution of 2.5 and 3.5 mL of 6% cellulose sulfate gel. This potential vaginal microbicide was mixed with dilute gadolinium chelate and inserted into the vagina. Magnetic resonance imaging was performed at baseline and repeated after 5, 20, 35 and 50 min with subjects either resting or ambulating between data acquisitions. Use of 2.5 mL with no ambulation gave low linear spread (53.1% of vaginal length) and surface contact (61.7%) at 50 min. Using 3.5 mL of gel with ambulation improved linear spread (84.5%) and surface contact (85.9%) of the gel in the same time interval without significant leakage. Most linear spreading took place in the first 5 min after insertion, although lateral spreading continued to increase with time especially when ambulation took place. Leakage and gel was not a significant problem at either volume.