Perhaps these days we have to look at gambits that start a bit later than the usual examples to find the most correct ones. For example the Qc2/d5 gambit in the 4.g3 Queen's Indian must be considered correct, no?  And in the Anti-Moscow Gambit White is one or two pawns down for a long time. Both lines have been extremely popular on top level the last few years. I've seen discussions on this forum on whether White really has anything against the Botvinnik Semi-Slav (obviously a gambit) and White can counter that with a gambit sideline of his own; 9.exf6!? gxh4 10.Ne5.

The Grünfeld has several examples: 1.d4 Nf6 2.c4 g6 3.Nc3 d5 4.cxd5 Nxd5 5.e4 Nxc3 6.bxc3 Bg7 7.Nf3 c5 8.Rb1 0-0 9.Be2 cxd4 10.cxd4 Qa5+ 11.Bd2 Qxa2 and 9...Nc6 10.d5 Bxc3+!? 11.Bd2 Bxd2+ 12.Qxd2 Na5 13.h4 being one. It is so "correct" in fact that many Black players would avoid taking the pawn, particuarly in the second way.

Main lines like 
1.d4 Nf6 2.c4 g6 3.Nc3 d5 4.cxd5 Nxd5 5.e4 Nxc3 6. bxc3 Bg7 7.Bc4 c5 8.Ne2 Nc6 9.Be3 O-O 10.O-O Bg4 11.f3 Na5 12.Bd3 cxd4 13.cxd4 Be6 14.d5  Bxa1 15.Qxa1 f6 16.Qd4 Bf7 17.Bh6 Re8 18.Bb5 e5 19.Qf2 Re7 have been worked out to ridiculous lengths, but still at the moment White is an exchange down for very serious compensation.

Or take the Korchnoi Gambit in the French Tarrasch:
1. e4 e6 2. d4 d5 3. Nd2 Nf6 4. e5 Nfd7 5. Bd3 c5 6. c3 Nc6 7. Ngf3 Qb6 8. O-O cxd4 9. cxd4 Nxd4 10. Nxd4 Qxd4 11. Nf3. Does anyone seriously doubt that Black is the one trying to equalize here?

I could go on and on. Of course which of these lines are considered critical tries for advantage (equality if black) at any given time will vary. But Pavlovic' statement is simply and obviously wrong.

I've been reading through Milos Pavlovic's book, Fighting the Ruy Lopez, and I was struck with a comment he made in the introduction.  After crediting Frank Marshall with the authorship of the Marshall Attack and the Marshall Gambit (Semi-Slav), Pavlovic claims that "these are arguably the only real gambits in chess which are recognized as correct, and have stood the test of time."

It got me thinking.  Is he right or is this just hyperbole?  Obviously the nature of compensation in any gambit is speculative and turns on a series of factors, stemming from space, tempo, development, king safety, etc.  But over time, theory chips away at these less materials features of the game.  Even the once-mighty King's Gambit is generally regarded as not entirely correct today, and the Queen's Gambit is hardly a gambit.

I would prefer that this thread not devolve into a list of how one's pet gambit is alive and well if only everyone played it the "right way" and opponents were polite enough to reply "properly."  What interests me is:

Do material concerns dominate the nature of opening play to such an extent that it is exceptionally rare for long term concessions in the opening to result in theoretically correct play?  And, if this is the case, are Marshall's two gambits exceptions that prove the rule, or just two openings whose incorrectness has not yet been exploited?