Duplicate Identification Fields
The purpose of duplicate identification is to avoid creating duplicates in the database. The method is to identify those look the same or similar while they actually represent the same person from your point of view.
When you select field(s) to identify duplicates, they are combined together as a single key for record comparisons. Adding a field means adding an AND checking condition. Hence when two records (assume the same person with the same email address) just differ in one checking field (e.g. an empty value in a checking field), these two records will NOT be regarded as duplicates.
Always try to select one field only where its data is unique among contacts. (e.g. membership ID, email address, but NOT contact name as it is never unique).
You should select multiple fields for duplication identification only when there is no single field with unique data and when you know what you are doing.
Finally, be consistent with the choices of every import, otherwise duplicates may also appear because of the existence of old records.
To view more, visit Import A File - Step 6: Identify Duplicates.