Not necessarily. If the data is heavily clustered around a single value, the percentile collapses to that number.