on 08/04/09 at 18:08:35:
1.d4 f5 2.Nc3 Nf6 3.Bg5 e6 may be underestimated. After 4.e4 fxe4 5.Nxe4 (interestingly this can be reached also via the move-order 2.e4 fxe4 3.Nc3 Nf6 4.Bg5 e6) 5...Be7 6.Bxf6! Bxf6, I have never seen the claim that 7.h4 is such a fabulous move but it seems to be the most popular and has been played by Levitt among others so it must have some merit.
I think 7...Nc6 must be best. I seem to remember that Bellin (or was it Silman?) in one of his old books on the Dutch recommended a plan involving ...b6, ...Bb7, ...Qe7 and ...0-0-0 in a similar position. That could be worth considering here too.
Bellin (1990) [on page 79] mentions 6.Nxf6+ Bxf6 7.h4! [his exclam] then he gives a variation showing how bad 7...Bxg5 is, and a game with 7...0-0 (which is almost never a good idea in this line, Black should strive for queenside castling). Finally he gives 7...Qe7! [again his exclam] without any indications as to plans or variations.
Harding (1976) and Schwarz (1964) also quote Schneider's analysis of 7...Bxg5. Both think White will get the advantage.
On 7...Nc6 Schwarz gives 8.Qd2 with idea 0-0-0, while on 7...d6 he gives 8.Bd3!
I would prefer Bellin's move and follow it with ...Nc6, ...b6, ...Bb7 and 0-0-0 as mentioned above. As usual, an alternative plan is ...d6, Bd7, Nc6 and 0-0-0.
A warning: the position after 7.h4 occurred 29 times in Mega database 2006: +24=3-2 or 88% for White.
At least one never book on the Dutch ignores 6.Nxf6+ and only gives 6.Bxf6, which seems a bit careless.