What weakens the signal? Walls? Trees? Windows? People passing in the path?
No, it’s not line-of-sight - it’s radio, not infrared. And a decent wifi router is going to have pretty significant range - stick a router anywhere in an average-size house, and anything with a wifi card in the house is going to be able to pick it up, no problems. In fact, you’ll probably get a signal in your yard. And across the street - which is why you should make sure to password-protect your network.
The wifi is on a pole outside, maybe 200 feet away. I have a decent signal when my laptop is on the windowsill but not when I sit down. This is a few feet more, and I wonder why the signal gets weak so fast.
While it is true that WiFi will go thru walls I beleive that the signals are still technically line of site. That is, they can’t go over a hill into a valley. I could be wrong, I’m not a wifi tech or anything like that. But I do know that they are trying to wire the whole county that I live in to WiFi (for free!) 
and one of the technical challenges they face is getting signals out to remote areas of the county where there are no publically owned poles.
Some info on that project here.
According to a public meeting I went to on the above mentioned wireless project many homeowners would have to purchase low cost repeaters from the project in order to get a decent signal in their home. Unless of course they wanted to sit on the patio and surf (not so cool in the winter!). 
I can’t wait until they hook my neigherhood up (even with only 128K access for free), but it’s not until 2008. 
Realworld, normal B-G and even the latest N spec Wifi is going to be pretty weak 200 feet from the source even through open air, and a wall could easily block a useful signal at that point.
From my experience it seems that the signal gets weakend considerably when going through walls diagonally. Which makes sense, I think. If the signal is traveling perpendicular to the wall it goes through, say, ten inches of “stuff”. On a diagonal, that can be increase to a couple of feet.
Then again, I could be completely wrong about this.
MrExcellent:
Your statement is somewhat misleading.
Basically, all radio communications are constrained by line of sight. Anyone with experience in the industry will tell you that obstructions will degrade the signal, often to the point that it is unintelligent and therefor useless. All other things being equal, an uninhibited path will produce better results than a path that does not have optimum line of sight.
Sure, you can sometimes get a signal through a wall, or get it to bounce off a window, but signal strength (and inherently, data quality) will suffer greatly.
This doesn’t take into account atmospheric effects such as ducting, of course, but these concepts are irrelevant to WiFi.
Then how can people get Europe on their ham radio?
Depends on the wall construction, and finish:
Studs with composition or vinyl siding outside and drywall will pass most of the signal.
Aluminum siding, not so much, and brick will attenuate a lot of signal as well.
Foil faced insulation, and the chicken wire or expanded metal used with stucco will also kill the signal.
The signals are still line of sight. They bounce from the ground up to the ionsphere and back down to earth see howstuffworks.com
I’ve had a great deal of problems with plaster and lath construction. Presumably the plaster of days gone by has a significant content of lead, alum, or some other agent that gave it consistency. My router in the back room of my apartment has problems giving signal in the kitchen (3 rooms away through 2 plaster walls). Newer, and particularly “pre-N” and MIMO routers tend to give better performance at longer ranges, but it’s unclear whether they signficantly extend maximum range. (Max specificed range is around 30 meters, but I’ve definitely gotten better than 50 meters given the right conditions.)
Wifi is “line of sight”; that is to say, you don’t actually have to be in direct visual contact with the access point, but the signal strength is proportional to the distance to the router, all other things being equal. It doesn’t bounce off the upper atmosphere or anysuch.
Stranger
On the topic of Ham (Mmmm HAM)
Yay at long last a thread that I can contribute too. I work in marine communications for the RFA (The Royal Fleet Auxiliary, the Royal Navy’s freight and fuel service) There are three types of Radio waves. HAM radio waves propagate in certain layers of the Ionosphere (High Frequency) so on a basic level they bounce up and down, and travel a greater distance than VHF on say a small pleasure boat. VHF is essentially visual range only. Just reaching the Horizon (depending on hight of the antenna, as the waves do not propagate at all…
At my parents’ house, we had the cable modem and router on the second floor. I could pick up a signal on each floor(basement, first floor, second floor, and then the attic where I lived). I couldn’t pick up a signal out in the backyard.
-Lil
I recall reading in a book (“Miracle at Midway” by Gordon Prange, I think) that they discovered this effect, or at least unexpectedly saw it in action, during the months leading up to the Battle of Midway. Bomber pilots flying out of Midway Isle were picking up UHF signals from pilots flying around Pearl Harbor, which should have been well beyond the range of their radios. Once they figured out that the radio transmissions were bouncing off the upper atmosphere and coming down much farther away, they started being much more discrete about what they talked about on their radios lest the Japanese be tuning in.
I set up a small wireless router in my GF’s attic, and the signal disappeared to virtually nothing as I walked downstairs with a laptop, less than 5 metres away. Now the floor is concrete (presumably with steel reinforcement in it) and the stairs are open metal, but still I thought that was pretty poor.
If you model your wifi router as a point source emitting electromagnetic radiation in a spherical manner, you will see that the signal power at a point “r” is proportional to the cube of the distance from the router to you (r^3).
What this means is that the signal strength from the the router does not decrease linearly with distance. As you move further and further away, the signal strength will decrease in an increasingly rapid manner.
What this means is that even though you may get a good signal at 61 metres (200 feet) from the router, you may not get anything at 65 metres because the signal strength will have decreased by 21% [(65^3 / 61^3)-1].
(I hope I haven’t made a stupid mistake somewhere. )
As an exercise for the reader, find the signal strength attenuation if our wifi router emits radiation laterally in a donut shape 
Power density at r is proportional to the square of r, actually. Are you, perhaps, thinking of magnetic field strength, which does follow the inverse cube law?
Inversely proportional to the square of r?