That's a very difficult question to answer Luis.
The two simple parts are just;
1) Antenna Gain at each end
2) Simple formula "spaceloss" between the antennas
Then you add some extra factors for "longer paths"
3)Atmospheric Attenuation (higher frequency problem usually), and rain can really change the results alot.
4)Blockage by buildings and mountains etc.(calculable, but not too accurately)
5)Multi-path
6)Atmospheric fading for long paths(below 75 mhz typically), varies with location, time of day, etc.

Note on Multi-path: This is one of the trickiest and most difficult to estimate.
One example; transmission between sona buoys over water - using 802.11b. The communication length was measured (about 4 years ago by a friend of mine in California) at 25% of the expected calculated length due to multipath bounce off the waves, 25% is equivalent to 24 dB more loss in the original pathloss calculation, so multi-path is huge in calculating effective path loss). Sending a simple signal instead of high data rate signals wouldn't have this added 24 dB though, so it does depend on what you're sending.
kch
Short paths with low data rates are easiest to calculate.

And of course the trasmit power--download rfprop it does it all

The path is calculated as Higgler said but the path loss figures give a probability of achieving a certain S/N ratio.  This is because atmospheric refraction and losses change and rainfall must be allowed for.  This usually means that over a long path you need space diversity in antennas to minimise signal dropout due to these effects.