It depends. Often it's faster just to keep the desired rows rather than delete all the undesired. That is, copy the good rows to another table, trunc the table, copy the good rows back in.
Otherwise you can DELETE in batches as Sean suggests ... but don't forget the filter on the DELETE statement itself :-).
SQL DBA,SQL Server MVP(07, 08, 09) A socialist is someone who will give you the shirt off *someone else's* back.