Grand Prix Attack 3. f4 or Bb5

In his Starting Out book on the GPA, GM Gawain Jones writes the majority of his book describing 1. e4 c5 2. Nc3 Nc6 3. f4. However, later in the book he says he's used 3. Bb5 for years, and when I check it out in ChessBase for the past 10 years or so, Bb5 is used slightly more often (about 3000 games vs 2300), and does significantly better (60% vs 47%). However, all of the videos I've come across still show f4. Any ideas what this is all about? If your stats disagree with mine, feel free to say so. Thanks in advance for any help.

Don't blindly follow DB statistics. What difference will someone else's stats make? Or to quote GM Fine quoting an amateur's explanation of a botched endgame to a master: "What good is the book if you don't know it and your opponent doesn't play it?"  (substitute "database" for "book" here)

No matter what stats you find in a DB they're going to depend on the size of the DB and if it's master games only or not, as well as how well it's been "weeded" for duplicate games, GM draws and plain old typos. The DB here is about 1.5 million games, my ChessBase BIG DB 2013 is now about 5.5 million games. But I don't think the stats it has are anywhere near as important as learning how to play whatever opening you like and deciding which openings and variations suit your style, temperament, ability etc.

In general I'd recommend playing whatever your pet line is from both sides of the bd whenever possible in order to learn it properly. Forget about winning statistics being the Alpha and Omega of your opening preparation and start figuring out your own strengthes and weaknesses in openings: do you like to attack or defend? Play open, semi-open or closed positions? Do you like to play gambits or accept them? Etc. 

GM Kaidanov has an introduction to his c3 Alapin Sicilian video in which he discusses the GPA and why he no longer recommends it to his students.  It's less than 10 minutes long and well worth checking out.  Apparently current theory leaves black with the bishop pair and an open game, not what white had intended at all.

I've seen 2.Nc3/3.Bb5 called the "Tiviakov Grand-Prix". Doesn't the book "Dangerous Weapons: Anti-Sicilians" have like five chapters on this line?

Edit: no, the chapters are in fact in "Experts on the Anti-Sicilian", and there are eight, not five

The only real difference is white is avoiding the Rossolimo with f4.. if you don't care about that then play either one.

Wow! Lots to look through here. Thanks, everyone.

I would like to mention that 3.Bb5 is known as the Tiviakov Grand Prix, not as the Rossolimo. There 's a pretty big difference. Since in Rossolimo, there is knight on f3 while in 3.Bb5 Grand Prix, there is a knight on c3 which allows f4. Personally I think 3.Bb5 is a better line to play than 3.f4 since it isn't that hard for Black to get equality.

Another reason I like Bb5 a lot more is that a lot of people don't know this and may think it will transpose to Grand Prix 3.f4 (Bb5) lines but it really won't. It's similar but the lines are little different. So White is very flexible on what he wants to play. White can also transpose back to Rossolimo with Nf3 if he wants. But that doesn't really happen often since one of the main moves against Bb5 is to play ...Nd4. 

Bb5 is also more popular in GM play than f4 from the last five years. So it shows that many GMs and top level players are employing 3.Bb5 which shows how sound Tiviakov Grand Prix is.

3.Bb5 Nd4 4.Nf3 is a common Rossolimo line that usually is reached with 2.Nf3 and 4.Nc3

I play 3.BB5 :D. I have good results with it. 

I only play Bb5 against 2...Nc6 to avoid any Sveshnikov

Gawain Jones gives the line 1.e4 c5 2..Nc3 Nc6 3.Bb5 Nd4 4.Bc4 e6  but in Beating the Anti-Sicilians by Kotronias he gives 4....a6.  Gawain Jones doesn't even mention that move.