Is it true that when performing 410(B) testing, one should not include employees only receiving a top-heavy contribution? Is it also true that if the 410(B) test fails without employees only receiving a top-heavy contribution, then those employees need to "benefit" at the same percent as everyone else (not just the 3% top-heavy)?

If 410(B) fails when treating those who only get a top heavy contribution as not benefitting, you don't necessarily have to provide the same benefit to these participants. You simply are no longer entitled to use the 401(a)(4) safe harbor for nondiscrimination, since the uniformity rule would not be satisfied. There may be other ways to pass 401(a)(4). See Reg. 1.401(a)(4)-2(B)(4).

In other words:
simply 'pretend' your plan is a cross tested plan with two groups, group A gets the full contribution, and group B gets 3%.
Test under the cross tested rules.

Can anyone provide a citation for the interpretation that a top heavy minimum contribution does not equal "benefitting" for 410(B) purposes? Thanks for your help.

I would be carefull to review your document. Most of the documents I deal with specify that the plan must pass 410(B) using the ratio percentage test, unless they were specifically designed to be allowed to pass the average benefits test.

Lynn Campbell, top heavy minimum contributions do equal "benefiting." As JAREL and Tom Poje point out, a problem can arise if the plan is using the design based safe harbor to pass 401(a)(4).

Under the safe harbor rules for 401(a)(4), there is an exception to the uniformity requirement for contributions, if the plan has multiple contribution formulas and one is the top heavy formula. But this exception to the uniformity requirement is available only if the plan can pass 410(B) coverage by treating an employee as not benefiting if his only allocation is the top heavy minimum.

If the plan doesn't pass 410(B) by ignoring the top heavy contribution, it has not failed 410(B); it has failed the safe harbor 401(a)(4) rules.

As prior posts point out, the plan may use other methods to pass 401(a)(4) non-discrimination; it just can not rely on the safe harbor.

see 1.401(a)(4)-2(B)(4)(vi)(D)(3)

'...the preceding sentance does not apply, however, unless the plan would satisfy section 410(B) if all employees who are benefiting under the plan solely as a result of receiving allocations under the top-heavy formula were treated as NOT CURRENTLY BENEFITING under the plan'

emphasis mine

i have a different scenario:
if an HCE were to show as not benefitting solely b/c he is getting only 3% th min, the 410b passes(doc states to pass coverage).

1 hce/key -full alloc
2 nhce --full alloc
1 hce/nonkey ---only th min
1 nhce --term $0

using the unsafe harbor this fails coverage, but using safe harbor it will pass(or am i incorrect?)
can i rely on the "safe harbor test" ?

ah, 2b or not 2b, that is the question or at least 1.401(a)(4)-2b

you did not indicate if the NHCE had less than or more than 500 hours.
I will assume > 500 hours, and so must be treated as includable and not benefiting.

Therefore, for the Schedule T you have
2/3 / 2/2 or 67%, fails ratio %
need to check avg ben test.
since ratio % > 50%, plan passes nondiscrimination classification test
avg ben % test should pass using allocation test. people get so used to using accruals for testing, and miss the obvious!

Now, the question is, do you have to perform the nondiscrim test because of the top heavy issue?
see 1.401(a)(4)-2(B)(4)(vi)(D)(3)
if you were to treat the HCE as includable and not benefiting then you have
2/3 / 1/2 = 133% which passes so no further testing required.
I also suspect you could have also used
1.401(a)(4)-2(B)(4)(v) - lower allocation for HCEs, instead