As I said earlier, I think many people confuse recursion itself with the way recursion is implemented with a given programming language, a given compiler and a given underlying platform.
I agree with Bruce. Recursion can be implemented in various ways. As I also said before, in some programming languages, a number of things are just NOT doable unless you use "recursion", and they don't necessarily use simple stacks either.
BUT - it is also true that in some prog. languages, especially those commonly used in the embedded world (such as C), recursion is most often implemented with the "stack", unless, as Bruce mentioned, it is just tail recursion, in which case decent modern compilers will optimize that and yield an iterative version. But it still depends on the optimization level, and still, less obvious recursion patterns are likely not to be optimized and still use the "stack". Then, whether this is a problem or not really depends on the underlying platform. But admittedly, it is pretty common with the typical C "memory model" that the "stack" be (often much) more limited in size than the "heap", and furthermore used both for data AND as a call stack (which IMO is a disastrous design choice, leading to a gigantic number of potential dangerous bugs and exploits, but that's another story), hence all the cautiousness with using recursion.