Related to the question Does $\zeta(3)$ have a connection with $\pi$?:
It is well known that
$$\zeta(2n) = f(2n) \pi^{2n}$$
where $f(n)$ is an function in rationals: (the denominator = OEIS A002432: 6, 90, 945, 9450).
Apery proved that $\zeta(3)$ is irrational, and presumably $f(2n+1)$ is so for all (but the latter is open).
But what else is known about $f(2n+1)$? What sort of pattern does it share (if any) with $f(2n)$? Numerically does it fit well? Is there any clue/progress since Apery as to its form? Of course, $f(2n+1)$ may in effect divide out the $\pi$, meaning that somehow $\pi$ is not involved in $\zeta(2n+1)$ in any essential way.
As Qiaochu writes in his comment, $\zeta(2n+1)$ (for $n \geq 1$) is expected to be algebraically independent of $\pi$, and furthermore for different values of $n$ these numbers are also expected to be algebraically independent of one another.
These expectations are far from being proved (as far as I know), but are not idle speculations: contemporary number theorists have a very tightly woven web of conjectures about values of $\zeta$-functions, and $L$-functions, which is supported by large amounts of theoretical and experimental evidence, and I don't think there is any reason to doubt that the expectations are correct.
Given this, I don't expect that the numbers $f(2n+1)$ will follow any significant pattern.