This is similar to the experience they had over at Salon. This was one of my favorite places to get news until they put up a pay wall, and in December they talked about how it hard hurt their traffic.
This is a great read, for people who actually care about the discussion of pay walls vs free.
http://www.guardian.co.uk/media/2009/dec/03/memories-paywall-pioneer
It "worked" for us in that it provided some revenue for Salon to survive through the leanest period of its existence. (We'd already completed the latest of three rounds of layoffs, and the entire staff took pay cuts, three weeks before 9/11.) But within a few months, as advertisers began dipping their toes back in the water and the influx of new subscribers who'd flocked to help us out in a crisis dwindled, we could see that the subscription model didn't provide much room for growth. So we tried something new: we put up an ad over the front door of the site. Subscribers wouldn't see it at all; other readers had to watch a 30-second video ad, then they got a "day pass."
The day pass approach was beloved by the advertisers and hated by many, though not all, readers. More important, by this point the public was, understandably, thoroughly confused about how to get to read Salon content. It took many years for our traffic to begin to grow again. Paywalls are psychological as much as navigational, and it's a lot easier to put them up than to take them down. Once web users get it in their head that your site is "closed" to them, if you ever change your mind and want them to come back, it's extremely difficult to get that word out.