In reply to ade sheffield:
For multipitch climbing in the U.S. (40-60m pitches, no bolted belays), my standard rack is a set of nuts, usually a set of micronuts, double cams from micros up to green Camalot, and one each of red, yellow, and blue Camalot. This has to be adjusted (in some cases radically) for certain types of climbing that involve far more big cracks.
I too have done many long routes using few or no cam placements, but those routes were two or more grades below my onsight limit at the time, so running it out was not a problem.
I think cam usage falls into three categories: (1) Essential in the sense that they radically diminish the risk level, (2) Not essential in previous sense, but they speed up climbing while reducing fatigue, (3) genuinely not needed. Of course, the spectrum is really continuous, not discrete.
For me, the "genuinely not needed" category is less than 20% of the pitches I encounter. Even when almost every placement is passive, a single cam can make a substantial difference in the risk level, and there is a night-and-day difference in the "endurance tax" between the exhaustion of fiddling opposed nuts into a horizontal crack at the lip of a roof and the relaxation of just stuffing in a cam. Although climbing grades don't take the difficulty of placing protection into account, having cams on strenuous routes can make all the difference in terms of success or failure.
Personally, I climbed for a solid ten years using only passive protection. In the areas I climb in the most, cams completely changed the risk level and the speed/fatigue level. Nowadays, the safety and difficulty of those climbs is rated assuming the leader has modern equipment, and a climber using only passive gear will most likely have an entirely different experience from anything suggested by the ratings.