I tend to think of each in this way:
Access: Great data modelling tool, good for single user data sets.
MySQL: SQL interface to an ISAM db layer. Good for storing data that is mostly read, with no need for referential integrity (i.e. fit for content management, but not financial or CRM type data)
Interbase: Big step up from MySQL in dbms functionality (and runs on windows) and with very good referential integrity, but still better in a read mostly environment with few committers.
Postgresql: Open Source dbms grows up. With the latest versions we have everything here except clustering / failover. Fully implements SQL92 (except for two statements that no seems to even ask for or completely understand... :-) Very fast under heavy parallel load, even with many writers. Write ahead log design now ensures complete recoverability should power be lost during operation.
Oracle, Informix, SAPdb: What the big sites run. Often handling thousands of users at a time or more. Expensive but not as expensive as the data it often protects...
MSSQL: In the same league as interbase or Sybase, has many features, but isn't know for enterprise class scalability or reliability. When properly managed and placed on a big piece of intel iron though, it's quite capable of handling a few hundred users at a time reliably.