A Percentile is simply the percentage of people in a User Group who are smaller than the specified size in the specific dimension. Normally, percentiles refer to specific dimensions, not whole people. For clarity PeopleSize refers to them as Dimension Percentiles.
It is recommended that you only set your percentiles after thinking carefully about the issues. The key issue is what happens to your excluded users - for example if it is a gradual non-serious loss of comfort then a moderate target like 2.5th to 97.5th might be the most suitable, however if health or safety issues arise then you have to set a much more extreme target like 1st to 99th or .5th to 99.5th. Normally you also need to know the cost of designing for the more extreme ranges before you can make an informed decision on design sizing.
It is always an advantage for a design to be tolerant of non-ideal sizing, rather than suddenly "not fit" a user who is a bit smaller or larger than the formal range. Examples of tolerant design are radiussed rather than sharp edges, generic/organic rather than highly fitted shapes, and adjustability. Many items have a limit at one end of the size range.
A Design Percentile is the percentage of people in a defined User Group who are smaller in EACH of the specified key dimensions. Click here for an explanation of the difference.
To give the percentage of the User Group accommodated by a whole design, PeopleSize subtracts the Small User Design Percentile from that of the Large User. Thus 3rd to 97th Design Percentile accommodates (97-3=) 94% of users altogether. This is the same as is normally done with dimension percentiles which refer to a single population. However note that with a single User Group 5th to 95th percentile only include 90% of the group, so the equivalent of 5th Female to 95th Male is 2.5th to 97.5th NOT 5th to 95th.
Measurements TO percentiles