I'd vote for picking something less common than Towers of Hanoi. When I had my first interview out of college in 1990, the engineer pointed me at a white board and asked me to write ToH and I answer with disdain, "recursive or non-recursive". It's that easy.
Hermite polynomials are a fairly good one to pick since it is a double recursive relationship.
H0 = 1
H1 = 2 * x
Hn+1(x) = 2 * x * Hn(x) - 2 * n * Hn-1(x)
(from here)
I had this as a CS101 assignment in Pascal, written as recursive, non-recursive, and iterative. I might still have my assignment book at home. If I do, I'll post the assignment.
EDIT - as promised - here's the text from the original assignment:
In this assignment you will write three programs that compute Hermite polynomials. The nth Hermite polynomial is x is denoted by H(x, n), where n is a non-negative integer and x is a real number. The Hermite polynomials satisfy the following recursive definition:
H(x, 0) = 1
H(x, 1) = 2 * x
H(x, n) = 2 * x * H(x, n-1) - 2 * (n - 1) * H(x, n - 2) for n > 1
Each program should contain a function H that computes the value of H(x, n). The programs should all prompt the user to enter values for x and n, and print out H(x, n) until he or she enters a negative value for n.
The first program, Herm1, should implement H as a recursive function (this is very easy). The second program, Herm2, should not use recursion but should use a stack to implement the recursive style of the first program. Notice that the computation of H is like that of Fibonacci; that is the number of recursive calls (or pushes on the stack) grows exponentially with n. For the third program, Herm3, you are to implement H with a loop that is traversed at most n times to compute H(x, n). Note: you may not assume any upper bound on n.
This assignment is worth 25 points.
Were I to ask this question, I'd probably ask for any implementation and then start in on questions about complexity and see if the candidate can get to this problem creating a stack issue on their own, and then start down the road of "and how would you solve that?"