TechCruch offers an interesting guest post from the developer of the iPhone game app Galaxay Impact. It's a case study comparing response to the app when it was free vs. how it was received when the company charged $.99 for the download. The company also used AdMob to promote and monetize the app as well.

[T]he free downloads vs for-fee downloads is about 400:1. That means for 220,000 downloads, our revenue amounted to $550. It’s obvious that there was no way we could make money out of this with a $.99 list price.

Another lesson learned: before the price change (from free to $.99), average downloads per day was above 10,000 but after price changed back to free, the average rebounced to about 1,000 per day, which continued for a long time. If we had not experimented with charging for the app, the total number of downloads would have been much higher.

Next, we decided to try advertising and updated Galaxy Impact with ads from AdMob along with other new features. There was a huge spike of update downloads with a 30,647 peak of November 22, two days after the update release.

Regarding ad revenue the company said that it's highest daily ad revenue was $16.37 and averages "about $2.50 per day." The company also said that updates didn't really expand the audience even though they increased traffic when they were released. 

The published takeaways were the following: 

• Free apps have a dramatically larger response
• Ad supported apps can make more money over the long term than charging for the download
• The overall ad revenues don't constitute a sustainable business

Caveats:

Many more such case studies need to be evaluated before any firm rules (or conclusions) such as these can be established definitively. For example, Pinch Media came to different conclusions from taking a look at the performance of a range of iPhone apps, suggesting making a paid app first and, if it succeeds, converting it into an ad-supported app. 

Ads and paid downloads are not incompatible; it's not necessarily "either-or" but can be "and."