Registration is necessary so that elections officials can verify that a person lives in the jurisdiction they are voting in - no carpetbagging, even to get the wack-o fundy stealth candidate off your neighbor's school board.  It's also needed to make sure that persons don't vote twice.

Many states require re-registrations as a way to keep the voter rolls from becoming clogged with dead folks or people who moved.  For example, Ohio recently adopted rules that if you have not voted in four years, you must respond via postcard to the voting registrar, or re-register.  I'm concerned that a lof of folks who registered more than four years ago but have not voted since then will suddenly find on November 2 that they needed to re-register, but that the deadline already passed.

Quoting from this site:

So, as a public service, here's how they do it. Let's start with Gallup. According to David Moore of Gallup:

Gallup asks each [RV] respondent seven LV screening questions, and gives each person an LV score of 0 to 7. [Assuming a turnout of 55 percent], the top 55% are classified as likely voters.
In practice that typically means all of the "7"s--given full weight--plus some proportion of those with lower scores (usually the "6"s), who are weighted down so that the size of the likely voter sample matches the projected turnout for the year (apparently 55 percent this year). All other voters are discarded from the sample.

What are the Gallup likely voter questions? Unfortunately, the exact questions and their wording are not released by Gallup along with their polling data, but the questions apparently involve past voting behavior, interest in the election, intention to vote in the election and knowledge of things like the location of the local polling place.

Another article about this:

http://slate.msn.com/id/2104727/%5C

Quoting from that article:

Some pollsters, for example, will ask only whether the voter showed up for the last election and whether they plan on doing so again. Since many people feel slightly guilty about not voting, or want to appear more politically engaged than they really are, such polls tend to dramatically overstate the number of likely voters. Pollsters refer to this approach as a "soft screen."

Other pollsters ask their respondents to qualify their answers, instead of giving simple yes-or-no replies. A typical question in such a poll might be: "Are you very likely to vote, somewhat likely, or not likely?" The answers are then weighted in order to account for the assumption that those answering "somewhat likely" will probably vote in fewer numbers than those responding "very likely." Pollsters may also ask those surveyed to rank their likelihood of voting on a scale of one to 10.

The most thorough polls ask whether respondents know where their polling places are and how they plan to travel there on Election Day. Another tactic is to couch the questions about past voting habits in soothing language, so the respondents don't feel as if an "I didn't vote last election" reply is equivalent to confessing that they're bad Americans. A popular way to phrase the voting-history question, for example, is along the lines of: "In the last election, did something come up that prevented you from voting?"