Is the following a correct understanding of nondiscrimination testing for controlled groups?

1) All employees of the controlled group will be treated as employees of one employer for qualified plan purposes.

2) If different employers within the controlled group have different plans, each covering only their own employees, then:

1. If each plan passes 410(B) coverage testing, then the other nondiscrimination tests (ADP,ACP,401(a)(4)) can apply to each plan separately.

2. If each plan cannot pass 410(B) coverage testing separately, then the plans will be treated as component parts of the same plan, and must pass the other nondiscrimination tests (ADP,ACP, 401(a)(4)) combining all participants from both plans.

What else is important when doing nondiscrimination and other qualified plan testing for controlled groups?

I think your comments are generally true, but watch for the 410(B) test. If each plan does not satisfy the ratio/percentage test, and must use the average benefits test, then you've got to combine plans for the average benefits test.

Also, watch for 401(a)(26) if there are any DB plans involved.

in regards to #1...'If each plan passes 410(B) coverage testing' this implies you have tested 410(B) separately, therefore '(ADP,ACP,401(a)(4)) can apply to each plan separately.' NO, those tests MUST be applied separately.
also, when you say each plan passes 410(B) separately, I assume you mean that you look at 'one plan' and are treating all other participants as 'includable and not benefitting' (410b testing not ADP testing, of course.
this leads to #2. my understanding would be as follows:
suppose i have 4 plans (members of a controlled group)
I can mix and match, rather than test all separatley or test all combined. again, whatever I combine for testing in 410(B) I combine for ADP testing.
so I could have 1,2, and 3 combined and 4 by itself or 1, 2 and 4 and 3 by itself, or 1 and 4 combined 2 by itself, 3 by itself, etc.
I think of it this way: controlled group, everything is combined. now I start setting up component plans, and those not included are treated as includable and not benefiting

A follow-up question....
The controlled group passes coverage any way we decide to split it (all four together passes, all four separately passes, etc.). However, the ADP test fails when aggregated together. When tested separately, Plans A,B and C pass but Plan D fails to the tune of $10K.
We've combined plans in an attempt to get the controlled group to pass (e.g., test A&D together, test B&D together, test B,C&D together, etc.). The best scenario is when Plans A&D are tested together... it still fails, but the excess amounts total $5K (as compared to the $10K Plan D would distribute if tested separately). It would seem that the second option is the wise choice, however, instead of distributing excess contributions to the HCEs of only Plan D, we are now distributing excess contributions to HCEs from both plans. Is it fair to 'drag' the HCEs from Plan A down with the HCEs from Plan D?

My question is: do we perform the tests separately and remove the $10K from Plan D (limit the # of participants receiving excess contributions) - OR - perform the tests by combining Plans A&D in an effort to reduce the amount of the excesses?

I'd present the options to the client(s) and let them duke it out. But, first make sure they can be aggregated. They need to have the same plan year ends and virtually the same benefits, rights, and features, or you'll have to test those within each plan and it may fail.

So, make sure aggregating them is an option first, then I'd leave it up to the client(s) to decide.