The only caveat with Bloom Filters is that they only answer the question of set membership with high probability. That is, they can be wrong (false positives). For almost all typical cases, you'd be better off with a binary hash (or even a normal hash table) so that you always get the right answer.