Wikipedia's entry on Najdorf Sicialian, 6.Bg5 mainline section

Below is a link to the wiki on the Sicilian Najdorf, 6.Bg5 mainline section. In studying the Najdorf with a view to playing it regularly, I've found this approach by White troubling, at least if I were ever to face a strong opponent with a good memory who was willing to draw when I might badly need to win. So far one of the most interesting approaches I've seen for Black with that in view may be the little-played 6...e6 7.f4 Be7 8.Qf3 Nc6 (inserting ...h6 and Bh4 is thought a bit less good by my database's computer evaluation, though it's scored a little better in terms of wins for Black, too, afaik[!]). Judith Polgar once played it unsuccessfully as Black in a high-level Active game. Aside from that, I've not quite rejected a number of lines for Black that might in some cases lead to so-called living draws if White wants, i.e. rather dull positions where there may still be hope to win in the end. An example would be 7...Qb6 8.Qd2 Nc6 9.0-0-0 Qxd4 10.Qxd4 Nxd4 11.Rxd4 Be7 [best hope objectively, of many moves tried]; here's the wiki I referred to, which is a bit short on specifics/depth:

https://en.wikipedia.org/wiki/Sicili...n_line:_6._Bg5

[edit: Two lines I've studied a bit with the help of CA16 database and Houdini may be of interest, as they are Black choices for certain books on the Najdorf relatively recently. One is the Poisoned Pawn Deferred, which goes 6...e6 7.f4 h6 8.Bh4 Qb6 and now best seems 9.a3 - then Black has a number of tries, but against the better lines for Black here, Houdini seems to eventually find forced draw(s) for White (if he wishes) when I looked. The second case is 6...Nbd7 when I think 7.f4 is critical - then 7...Qc7 is the modern choice, after which 8.Qf3 g6 (again the modern way) 9.0-0-0 intending Nd5 is at least solid looking for White in the event of ...Nxd5 exd5 when ...b5 can allow further simplification by Nc6 at some point. However, I think more interesting, if White is going for advantage, may be the relatively little-tested 8.Bd3, possibly followed by Qe2 and 0-0-0, partly with the hope that in some cases Qe2 in one move might save a useful tempo over some Najdorf lines with Qf3 and ...e6 played, if White eventually wants to play Qe2.]

Two other lines I found interesting for Black when I studied 6.Bg5 a while back are 6...e6 7.f4 Qc7 (trendy for a while) and the surprisingly common 6...Nc6, transposing to a Richter-Rauzer where Black has played ...a6 early (but this might easily transpose to major lines of the Richter-Rauzer where ...a6 is included). Why not play a Classical Sicilian instead of the Najdorf? Well one possible reason is a Sozin setup (6.Bc4) is more potent against that than in the Scheveningen (move order for that is commonly with an early ...e6) via the Najdorf move-order. In the Classical, with both Black Kts out, 6.Bc4 Qb6 can be very complicated, and allowing e.g. a Velimirovic Attack by transposing to a Scheveningen with ...Nc6 included is also complicated (and dangerous). Another reason is that people who like to play the Najdorf need to transpose out of it in case of 6.Bc4 e6 (a Scheveningen), anyway (though Black is fairly happy in that case these days since he's not committed to ...Nc6).