A record in an NCBI database is designated with a number, called a unique record identifier (UID), that unambiguously differentiates the record from all other records in the database. In general, if you select and search a database from the All Databases menu on the NCBI home page, you will obtain your results in one of NCBI's default summary display formats. You will then be able to locate the UID for a given record at the bottom of its summary.
A UID is listed is:
“XX” signifies additional characters that are often added to associate the UID with its corresponding database. For example, PMID is the UID for the PubMed database, CID is for the PubChem Compound database, MMDB ID for the Structure (NCBI's Molecular Modeling Database), and so on. The Nucleotide (including EST and GSS) and Protein databases have historically used UIDs called GI numbers, but NCBI is in the process of transitioning to accession numbers instead.
If you have recorded the UID for a particular record, you can retrieve that record in the future by using the UID as your search term. Enter the number only. Using UIDs, you can also establish links to individual records or refer to the records in publications. For example, referring to a gene by its Entrez Gene ID in a publication is not ambiguous while using symbols that can be shared between different genes is. Database and software developers often base their programming applications on UIDs, and the E-utilities (the Entrez API) currently use UIDs as their primary input/output keys.
UIDs are not the only unique identifiers that NCBI staff (or our collaborators) assign to the records/data. Among other unique identifiers, you will often encounter accession numbers.