Geoff on the right track about the way Suzuki groups occur naturally as subgroups of others.   But in view of the incomplete formulation of the original question, and the string of comments following (with some references not directly relevant to Suzuki groups), it's worth pointing out sources in the literature.    

1) In his 1962 *Annals of Mathematics* paper, Suzuki was studying special 2-transitive permutation groups (not yet in the Lie context), but already in that paper he worked out explicitly the limited types of possible subgroups which can occur in his new simple groups.   These include certain smaller Suzuki groups.

2) In Carter's book *Simple Groups of Lie Type* (1968) and in Steinberg's 1967-68 Yale lectures on Chevalley groups (the lattter notes available online), the Chevalley groups and twisted groups of types $A, D, E_6$ along with the groups of Suzuki and Ree, were studied in a unified way but with different notational schemes.    For more technical treatment, see also Number 3 in the AMS book series *The Classification of the Finite Simple Groups* by Gorenstein-Lyons-Solomon.  Their notation for Suzuki groups (Chapter 2) differs from others.   But in all these sources the group involves an odd power $q^2$ of 2, and the group order is given as $q^4 (q^2-1)(q^4+1).$   This convention is not the same as Peter's but is useful for comparisons with a related Chevalley group order.   The Suzuki groups themselves are denoted in the Lie theory setting by 
$^2\!B_2(q)$ or such.

3) The book by G-L-S then formulates Suzuki's subgroup theorem in their Theorem 6.5.4.   Here the criterion for one Suzuki group to occur as a subgroup of another one is that its odd exponent of 2 properly divide the odd exponent for the larger group.  This is essentially Geoff's observation, confirming the numerical observation made in Peter's comment.   For instance, you can have the respective values $q=2^3$ and $q=2^9$.   (There are parallel inclusions of Suzuki groups in Chevalley groups of type $B_2$ and of Chevalley groups in each other, but not such direct inclusions relating all of the groups involved.)

4) The numerical divisibility results are a natural byproduct, but easy to observe directly as in Peter's comment.   Whether there are other "accidental" numerical divisibility possibilities for the group orders, I don't know.