Andy Wingo has written a somewhat opinionated guide to Scheme implementations.
It is somewhat unfortunate that the only implementations of Scheme that come with a standard IDE are Racket and Chez Scheme. I had a number of arguments with one prominent member of the Racket community, who once told me, "There cannot be two types of members in this community: those who contribute code, and those who don't." I am a haiku poet-aspirant who occasionally writes code for fun, and who works as a professional translator; as such, I simply don't have enough time or energy to contribute code; I prefer submitting many bug reports, instead.
Also, I don't really like the style of coding for many of the libraries in the Racket community. A few years ago, I read a blog entry by someone who commented that many Racket libraries tend to use code that is rather verbose and not particularly elegant, and one of them even went so far as to claim that Racket code was actually more bloated than Common Lisp code.
While I'm not certain whether I agree with that last assessment, personally, I prefer such implementations as Gauche Scheme and MIT Scheme, which more closely following the minimalist philosophy of R5RS Scheme, than Racket, which tends to follow the contrasting non-minimalist philosophy of R6RS Scheme.
Chez Scheme has a fast implementation, good documentation, and a friendly architect, but lacks mailing lists. Since my main goal in studying a programming language is to discuss it, although I enjoy programming in Chez Scheme, I have great difficulty in discussing it in a forum that is restricted to that implementation.
The main problem with Scheme has always been that there has never been a single main reference implementation, unlike, say, Haskell. Haskell has the GHC implementation, which is a reference implementation; unfortunately, Scheme does not have any corresponding reference implementation.
Although I have had a number of arguments in the past with some members in the Haskell community, nobody there has ever told me anything equivalent to "There cannot be two types of members in this community: those who contribute code, and those who don't." The worst that has happened is that I once got flamed on Haskell-Cafe by a French mathematics professor for trying to discuss a topic in mathematics without really knowing what I was talking about; because of his rather condescending reaction at the time, I later became extremely careful in discussing anything relating to mathematics with any Haskellite. However, his reaction does make sense for an elitist pedantic researcher in mathematics, and as a pedant myself, I can understand and somewhat appreciate how pedantic mathematicians think.