It all depends on what information you want to analyse.
If you're doing aggregations and rollups then you'd want to pull this data into a database for analysis. Pulling your data into a database will give you access to indexes and better querying tools.
If you're doing infrequent one-off simple queries then LogParser might be sufficient for your needs. However you'll be constantly scanning unindexed flat files looking for data which is I/O intensive.
But as you say, without knowing more about your specific scenario it's hard to say what would be best.