For every dozen Twitter followers you pick up, it’s likely at least one is a spam account. Finding and removing these spammy accounts can take a lot of time and attention.
Or you could use TwitBlock. It will analyze your follower list to help you find who’s real and who is a spambot.
It uses a Bayesian filter to look over all of the accounts that follow you and assign them a score. The higher the score, the more likely they are to be fake.
TwitBlock even maintains leaderboards so you can see the most overused fake avatars and how often the accounts that follow you are blocked.