Jordan Hölder decomposition for group objects

Is there some generalization of the Jordan-Hölder decomposition for group objects in a category $\mathcal{C}$?

If $\mathcal{C}$ is the category Sch$(S)$ of schemes over a base scheme $S$ then (I think) this is true, also probably for other categories of "spaces" like Top or Diff it should be true, but I don't have any idea for general categories.

-
In a general category, it's not a priori true that even the isomorphism theorems will hold. For an example, see this answer by JS Milne about classical reductive groups. If these "theorems" don't hold, I think it's pretty clear that you can't prove Jordan-Hölder. –  Harry Gindi Mar 20 '10 at 13:01
Could you be more specific as to what you mean by the Jordan-Holder decomposition? Do you mean the uniqueness of composition factors (as multisets) in any two composition series? (Of course any kind of statement about existence of composition series is not going to be true even for general group objects in the category of sets.) –  Pete L. Clark Mar 20 '10 at 16:12
That's exactly what I meant, given that there exists a composition series, it should be unique up to permutation. After thinkingof it for a while, I believe that it should be true for any category in which the second isomorphism theorem holds. –  Cristos A. Ruiz Mar 20 '10 at 17:02
Which one is the second iso thm? It varies according to who you ask. –  Tom Leinster Mar 21 '10 at 2:41
I mean the following: if $S$ is a subgroup and $N$ a normal subgroup of $G$ then $N$ is normal in $NS$, $N\cap S$ is normal in $S$ and $NS/N$ is isomorphic to $S/N\cap S$. –  Cristos A. Ruiz Mar 22 '10 at 18:34