You don't generally hear that you should develop for Android first, but even if you go iOS first, Android comes second. That's the traditional wisdom, anyways. According to Semil Shah on Haywire, though, "iOS first" is an understatement. It should be iOS first, and Android in the distant future, if at all. Shah is fairly direct with his point of view:

The most common trap here is the early iOS app which gets some buzz. All of a sudden, the founders hear “When are you building for Android?” The natural, enthusiastic response to sincere requests of the Android chorus is to go ahead and build for Android and seek more downloads, more growth, more revenue. I have a different view though. The proper response is: “No. Buy an iPhone.”

Shah's reasoning is presented in three central arguments:

• Android's fragmentation problem is too much for small teams
• iOS users have all the money (and their numbers are growing)
• Future Apple hardware (iPhone 5c, for example) may level the playing field

And these are some interesting thoughts. The third point in particular is one you don't hear often - one of the big selling points of Android as a development platform is the massive reach, which is a product of the sheer number of phones in consumers' hands. After all, iOS traditionally has a higher barrier of entry when it comes to economics. 

On the other hand, we've already seen some counterarguments to some of these points. For example, if you ask Nick Bradbury, fragmentation is a completely overblown problem, and according to Danny Roa, there's not really that much point to supporting older devices in the first place.

Similarly, Kevin Quach suggests that the common notions surrounding Android's monetization opportunities - that they're not there, basically, at least compared to iOS - are bunk as well. 

In other words, the "iOS first" vs. "Android first" argument may not be so clear in either direction. Check out Shah's full article for all the details.