I was polyphasic for about six months a number of years ago. My schedule was a fairly standard 6x20 minute naps (commonly called the Uberman schedule). After the initial (very, very difficult) adaptation period, I felt great nearly all of the time. I would get tired right before a nap, but by the end I didn't need an alarm to wake up after 20 minutes, and naps left me feeling completely refreshed and alert. I was in graduate school at the time, so my schedule was pretty flexible, and I had a mixture of cognitively demanding work (my research: a project in geophysical fluid dynamics) and less mentally demanding work (grading, family and home responsibilities). Again, after the adaptation period, I wasn't particularly cognitively impaired.
There is a lot of misinformation out on the web about polyphasic sleep, both by its proponents as well as its detractors. Some of the things that have been written (polyphasic sleepers only get REM, for example) are just false. One problem is that there isn't a very large body of peer reviewed literature on the subject; the canonical reference is a volume of conference proceedings from 1991 which can be hard to lay hands on. The fact that (a) adaptation takes a long time (on the order of weeks) and (b) the polyphasic schedule is incompatible with the way most people live their lives has made researching it difficult.
I'm currently involved in a polyphasic sleep experiment organized by the folks at Zeo. They asked for some volunteers to use their hardware to monitor our sleep as we either adapt to a polyphasic schedule or (for those already adapted) function on one. We're not following any kind of rigorous experimental protocol, so this isn't going to be publishable research, but it will at least provide us with some measured anecdotes.