Database entry - avoiding duplicate records

I'm working on a web application, and the application allows people to enter, edit, and search for people. I have one table that stores all people, and an autonumber is automatically assigned to a person when they are entered. Therefore, since I'm using an autonumber as a primary key, there is the possibility that a person can be duplicated.

I want to avoid duplicates as much as possible. I've thought of 2 solutions, but I'd really like to hear more. Here's what I have:

1. Don't provide the user with a link to the entry page until they've ran a search (i.e., the link to the entry page is on the search page).
2. After submission on the entry page and before the record is written, show a list of possible candidates for the information that was entered. If the person already exists, they can cancel the entry process.

What are some other techniques I can use to avoid duplicate names?

I'd go with something akin to option 2. Whenever a name is submitted, do a search, and if you come up with a match, offer the user the chance to select from the matches. If there is no collision, then enter a new record without further prompting.

And what do you call a duplicate person?
There's thousands of people with the same name, so that can't be it (at one stage there were 3 people with the exact same first AND last name working in the company that employs me, which at that stage was about 60 people total. Now that's extreme but you get the point).

And some people might have outdated information about someone, so they won't recognise that person when he's listed, leading to duplicate information anyway (2 entries for one person with a different address for example).

You can never fully prevent it, no matter what you do.