Here is some of the best information I have ever been able to find on the subject.

Glorfindel1235 wrote: '(...) What are the reasons for doubting the marginal note? Was extensive work done that included Balrogs after this note was written and did Tolkien continue to write about many Balrogs in those works?'

 

As I say I tend to lean toward the marginal note myself. Tolkien didn't write anything extensive on Balrogs after this note was written (that I know of anyway), and the doubt, or the measure of doubt, has only to do with the 'survival' of other citations. So far I've counted 4 surviving references. Here they are -- followed by the alterations made to The Silmarillion by Christopher Tolkien:

1: 'Wherefore each embassy came with greater force than was agreed, but Morgoth sent the greater, and they were Balrogs. Maidros was ambushed...' Of The Siege of Angband (Quenta Silmarillion) [] '... but Morgoth sent the more, and there were Balrogs.' Of The Return of the Noldor (The Silmarillion)

2: 'Sauron came against Orodreth, the warden of the tower, with a host of Balrogs.' Of the Ruin of Beleriand And the Fall of Fingolfin (Quenta Silmarillion) [] '... named Gorthaur, came against Orodreth, the warden of the tower upon Tol Sirion.' Of The Ruin Of Beleriand (The Silmarillion)

3: 'There came wolves and serpents, and there came Balrogs one thousand,...' Of the Fourth Battle: Nírnaith Arnediad (Quenta Silmarillion) [] 'There came wolves and wolfriders, and there came Balrogs, and dragons...' Of The Fifth Battle (The Silmarillion)

[1 This description (from the QS tradition) survived into LQS despite a number of other post Lord of the Rings revisions to this chapter. 2 The second example (Orodreth and etc) also was not revised -- with Tolkien even altering §143 of the chapter, but not the 'host' of Balrogs passage (see below). 3 The third example 'survived' too, but noting CJRT's description under The Last Chapters Of The Quenta Silmarillion, it looks like JRRT never really got around to truly revising this chapter in any case.]

4. The Grey Annals contains 'Balrogs a thousand' §230, but nothing is noted as to any changes in the later 1950s.

 

It's perfectly reasonable that these references survived simply because Tolkien missed them, or never got around to altering them. And as we can see, Christopher Tolkien's revisions alter these 'surviving' references to large numbers of Balrogs. 

 

As for chronology, what we have here is (basically) early 1950s and later 1950s. The Annals of Aman marginal note and revision are dated to the later 1950s, and if I read Christopher Tolkien's explanations correctly, we seem to have at least one other later correction of possible interest. First, in connection with the marginal note to Annals of Aman (which states the number), the actual revision to the text reads: 'a host of Balrogs, the last of his servants that remained' revised to 'his Balrogs, the last of his servants that remained faithful to him.'

So in this instance, the word 'host' was revised by JRRT himself. Moving from the Annals to Quenta Silmarillion (mid to later 1930s) 'But at length after the fall of Fingolfin, which is told hereafter, Sauron came against Orodreth, the warden of the tower, with a host of Balrogs. Sauron was the chief servant of the evil Vala, whom he had suborned to his service in Valinor from among the people of the gods. He was become a wizard...'

 

LQS (early 1950s) changes made in notes referencing the numbered paragraphs: 'Sauron was the chief servant of the evil Vala...' revised to 'Now Sauron, whom the Noldor called Gorthu, was the chief servant of Morgoth. In Valinor he had dwelt...' So the very next sentence after 'host of Balrogs' was changed, in a note, in the early 1950s. Next...

 

... LQS (later 1950s) the earlier changes were taken up into this later series of copies, thus, as I read things anyway, Tolkien is not merely looking at the note now, but looking at the changes made and put back into context with what he had written in the 1930s. And in the later 1950s Tolkien changes section 143 again: 'Now Sauron, whom the Noldor call Gorthu' is revised to 'Now Sauron, whom the Sindar call Gorthaur.' And JRRT also struck out the sentence: 'In Valinor he had dwelt...'

 

And again, the sentence before 'Now Sauron...' still included the very same word Tolkien either had corrected, or would correct, in Annals of Aman: 'with a host of Balrogs', and one wonders how he could have missed this reference, or at least 'missed it' if these corrections came after his new idea. 

 

The Annals of Aman note could still be later, despite being in the same general time frame as the corrections to LQS II; but we don't know that really -- or at least I don't know that, although it's possible I've missed something of course. Tolkien might have simply missed this reference in any case, even if this revision came after; but I don't think we know that either. 

 

I'll grant that, even if I'm correct, the measure of doubt that any of the above might raise is arguably small enough...

 

... but still 

Well, one could emphasize the matter various ways too:

Tolkien has a new idea in (or after) the late 1950s (at most seven Balrogs), notes it down on the text he is working with at the moment, and can't even remember where the other passages are, for correction, and means to get to them later; but never really does.

The Orodreth 'host of Balrogs' text is really the only instance I can point to, so far, where Tolkien did make a correction in the later 1950s (or after) -- and to a 'nearby description' (the very next sentence) -- leaving the preceding sentence alone.

Tolkien may have even seen 'host of Balrogs' there and forgot (just for the quick moment he was correcting the bit about Sauron) that he had drastically reduced Balrog numbers. Again, if in fact this LQS revision was later than the AAm revision.

Characterize it that way and it seems like very minor doubt -- on the other hand, all we really have is one marginal note that connects to one authorial revision, compared to decades of Tolkien imagining very many Balrogs existed! earlier decades yes; but still it was a long held idea, and maybe Tolkien wasn't quite so sure he was ready to give it up.

On the other other hand, all we need is one  LOL

So while HME may be detailed, there's still plenty of questions left if one wants to really niggle. How many people might quote other people quoting CJRT that 'Orcs from Men' appears to have been his father's final idea on the matter -- some of them not realizing that in HME the very next thing CJRT writes is (something like): but as always, it is not quite so easy! And here's an example where, IMO, the reasons CJRT points to (as to why Orcs from Men appearing as a final idea isn't so easy a conclusion) seem only to raise 'minimal doubt' to me; but he still raised them.

I understand that sometimes I confuse things. Maybe if I simplify the external history:

1916-ish to 1937-ish Tolkien imagines many Balrogs existing. In these years Balrogs are not Maiar, but seem to be powerful 'demons' in any case.

Now Tolkien takes a break (in general) from his Silmarillion, and poetry, to write The Lord of the Rings (sequel to The Hobbit), and after finishing it, in the early 1950s he goes back to writing and updating the Silmarillion, as he wants both works published together.

early 1950s if I recall correctly Tolkien begins to consider his Balrogs as Maiar in this general time frame (or Umaiar since they are bad guys) -- but still considers them to be numerous, as can be noted in his descriptions.

later 1950s at some point after Tolkien has had copies made of the 'Silmarillion stuff', he will take a copy of one of his Silmarillion-related works -- called the Annals of Aman -- and revise 'host' of Balrogs to 'his' Balrogs, also writing in the margin that there should not be more than three, or at most seven, ever existing. This is a notable revision, considering how much time has passed where Balrogs, no matter what they were or how they were created, could exist in enough numbers as to comprise a host.

The Balrogs are certainly Maiar by the later 1950s, and possibly (as some readers have conjectured) they are reimagined as few in number because Tolkien considers them too powerful to exist in great numbers. In any case, Tolkien will not however, and for whatever reason, correct every already existing reference to large numbers of Balrogs -- so they remained in the texts as they still do -- but Christopher Tolkien altered any surviving references in Quenta Silmarillion so that his 1977 Silmarillion does not refer to large numbers specifically. Possibly he does this because of his father's marginal note, and in any case the matter of actual numbers is left somewhat ambiguous for the published Silmarillion.

Ultimately Tolkien will never truly finish updating his Silmarillion, especially the later chapters; although he makes some cursory corrections to later parts.

And as I have pointed out, sometime in the later 1950s (or after) -- thus in the same very general time frame as the marginal note to Annals of Aman (where JRRT corrected 'host' of Balrogs) -- Tolkien corrects a statement from Quenta Silmarillion about Sauron but does not correct 'host of Balrogs' in the preceding sentence. Why not correct this reference in Quenta Silmarillion, just as he had corrected the one in Annals of Aman, if truly there could be no 'host' of Balrogs?

Possible and reasonable answers can be noted, but none can say with surety 

Well, I realize you're joking there, but just to note: if some of my posts are confusing it's my fault for not being clearer -- or perhaps the matter is simply detailed and confusing, especially for those who don't have HME -- as having HME makes it easier in any case, although I realize not everyone is into HME.

By simplifying the external chronology I didn't and don't mean to imply anything negative about anyone's understanding. 

Not that you said otherwise... but just to be clear

OK, no worries (as they say down under)

I'll also add that while the Annalistic traditions (The Annals of Aman and the Annals of Beleriand or 'Grey Annals') may be new to people who haven't read The History of Middle-Earth series, Christopher Tolkien actually used these sources in the 1977 Silmarillion, blending the Annals (Aman and Grey) with Quenta Silmarillion.

Christopher Tolkien felt that the Annals (traditions in which the years would be noted along with the major, or notable, events of that year), which in the 1930s were relatively brief, were becoming fuller in the early 1950s, and becoming less distinguishable in style from the Quenta Silmarillion tradition -- so as I say he blended them -- with the Tale of Years taking over for the Annals.

So Silmarillion readers have already read some of these texts, essentially.

Huan, I meant angmar. I believe your onto something either way.

You're onto extreme awesome-ness. The thread brings up another question: Had Gandalf not possessed Narya, the ring of fire, would he have fared so well in the battle against the Balrog? 'Long I fell, and he fell with me. His fire was about me. I was burned.'   Book III, chapter V, pg. 501, The White Rider

     Was the ring of fire the only reason he survived its clutches?
Do you think when the Wizards took up the mantle of Istari (or it was bestowed upon them) that it relieved them of their former Maiar prowess? Or do you think that they had the power of the Maiar but were so obedient that they simply didn't use it? 

In the books, I have only found evidence of minor spells, but one would think that Saruman would've utilized them more if he possessed them.