The fluoride release from a commercial, restorative glass ionomer cement was found to be strongly dependent on sample surface area rather than volume. This was noted for disc, cylindrical- and bar-shaped specimens over periods ranging from 1 day to 3 yr. Release from all shapes of specimen followed the established pattern of an initial non-linear region followed by one where release was proportional to the square root of time. If fluoride levels in the cement matrix of specimens were artificially increased by incorporation during the mixing then the release pattern during the first few months was altered. The initial release increased for some specimen sizes and decreased for others. The dependency on surface area was greatly reduced for several months. By the time a year had elapsed the correlation between fluoride ion release and surface area had been re-established. The influence of additional fluoride during setting can therefore act to perturb the normal release pattern and may in some instances reduce the initial fluoride release. Release should be quoted in terms of, or with measurements of, the surface area of specimens under investigation.