The easist way:

create table nodups as select distinct * from table_with_dups;

This may not work on really huge tables, but I've done it on a 1.5TB table on a 10400.  It won't work if you have timestamp columns or something which might be different between two rows.

I meant it won't work if the timestamps are different or if you have any differences at all in the records.  Sometimes you might have duplicates based off your primary key, but different timestamps or some other column like that.

This approach eliminates duplicates, but you need to be careful that the MAXed columns truly make business sense, like an arrival timestamp.  If you MAX a code column or foreign key that varies within your grouped primary key columns, then the results may be inaccurate.  The dupes are gone, but the problem is "swept under the carpet", so to speak.  Here is a suggestion to avoid inaccurate results:  1) Confirm the true primary key, 2) ensure that all non-additive columns exist in the Group By except the time stamp  3)  If there are still duplicates relative to the primary key, then get business criteria for which of the dupicates to preserve and which can be eliminated..