I worked with several of the MaxSonar EZ1 sensors back in 2007. I found that they generally worked very well, but did occasionally interfere if the sensors were parallel to each other and between roughly 1 and 2 feet apart. Unsurprisingly, there were also problems if the sensors were pointed directly at each other.
In general, the sensors worked very well, and I wouldn't worry much about interference. We were using them close together for sensing game pieces (and chose to use Sharp IR sensors instead) but for navigation-level sensing, the MaxSonar sensors are an excellent choice.
As with any sensor, you should filter the input, especially if a result is going to cause irreversible actions. For example, instead of driving until you get a reading less than 8 feet, drive until you get three successive readings under 8 feet. It's the real world, so expect a bit of noise in the readings!
These links might be useful:
http://maxbotix.com/performance.htm
http://maxbotix.com/articles/012.htm