Surprisingly, even though Murray sports a 6-4 advantage against Federer overall, Federer has won their 1(!) meeting in a major 5-set match (which was the 2008 final won by Federer in straight sets) and by some quirk, all their matches have been played on hard courts, although both players possess all-court games.
Murray has looked the sharpest of all the top players all tournament long and has only dropped a single set. Federer has been up and down but he was scintillating in his straight-set elimination of Tsonga in the semifinal. There's no question in my mind if Federer plays his best tennis, he will win the match. It's not clear at 28 years old, facing younger opponents like his arch-rival Nadal, the 21-year-old 6'6" phenom J uan Martín del Potro (who took Federer out in a 5th set US Open final last year) and the 22-year olds Djokovic and Murray how much longer it will be true that when Federer plays his best no one can beat him. I suspect we will find out this year if Federer can even still play his best tennis.
After correctly predicting the result of every single men's match of the quarterfinal and semifinal rounds except for Jo-Wilfried Tsonga's surprising 5-set dispatch of 2008 Australian Open champion Novak Djokovic, I find this match surprisingly difficult to predict. However, using a similar technique I deployed in predicting the result of the women's final, I will endeavor to approximate a probability of the winner of tonight's match.
There are 6 possibilities: Federer in 3 sets, Federer in 4 sets, Federer in 5 sets, Murray in 3 sets, Murray in 4 sets, Murray in 5 sets. If the match is 3 sets long, I think there is a 76% chance the winner is Federer (24% chance it is Murray). If the match is 4 sets long, there is a 55% chance of Federer winning (a 45% chance for Murray). If the match is 5 sets long, there is a 40% chance of Federer winning (a 60% chance for Murray). Crunching the numbers this corresponds to a 57% probability that Federer will win the match. [One key assumption of this method is that 3 set, 4 set and 5 set matches are equally likely. This is probably not true, if someone has the data on what percentage of Grand Slam matches are 3-set, 4-set or 5-set contests, that would greatly help me improve this technique. In the meantime the assumption of equal likelihood does massively impact the results.]
MadProfessah's pick: Federer in straight sets (Murray in 4 or 5 sets).