Welcome to PF, cyleung! It is pretty meaningless to talk about the the temperature of the vacuum. You must have some kind of thermometer, and no one has figured out how to make one without using matter. Until the photons crash into a hunk of matter, they are undetectable. But using 'thermometers' we can measure the photon density and energy per unit volume of empty space. And this works out to be the temperature equivalent of 2.725K, as measured by COBE and WMAP.
Indeed, as turbo noted, predictions of the temperature of space have been proposed since Stefan found, in 1879, that the radiation, F, emitted by a blackbody at temperature T is given by F=s x T^4, where s is the Stefan-Boltzmann’s constant - which was derived by Boltzmann in 1884. The earliest known estimate for the temperature of empty space was by Guillaume in 1896. Applying the SB formula to a crude estimate of the brightness of the night sky, he obtained a value of 5-6K. Eddington, in 1926, arrived at a similar result, 3.18K by similar means. But this method is fundamentally flawed. It only considers the effect due to stars in our own galaxy. Intergalactic space is millions of times more diffuse than a typical galaxy, hence the contribution of starlight to the CMB temperature in deep space is negligible. Here is a relevant article:
http://www.astro.ucla.edu/~wright/Eddington-T0.html
Note that in conducting the COBE and WMAP studies, the researchers subtracted out the contribution of our galaxy to obtain the actual CMB temperature.
The pre-CMB predictions of the temperature of space are sometimes cited by critics of big bang theory. But they do not hold water in the face of hard evidence. For example, in a static [steady state] universe, the background temperature would be constant at all times. But this notion was convincingly refuted in a paper by Varshalovich et al in 2000 when they measured the temperature of gasses in deep space billions of years ago. Here is a article on that:
http://www.eso.org/outreach/press-rel/pr-2000/pr-27-00.html