Months ago I found a comparative test that an Australian guy did. He tests CPU coolers and other hardware and publishes the results on his hardware and gadget review site. He used the same rig that he uses to test coolers - a electrically heated block which delivers a measured amount of power in Watts, containing a temperature probe, and another thermometer to measure the ambient air temperature. The difference between the two temperatures (the "delta-T") divided by the heat in Watts gives the thermal transference of the setup in °C/W
http://www.dansdata.com/goop.htmHe tried Arctic Silver 3 two ways, first in "More Is Better" style, putting on a half-millimetre layer of the stuff. "Lots of people over-apply thermal grease, and I wanted to see what difference it might make." Then he scraped most of it off and redid the test.
Vegemite is a kind of savoury (British & Commonwealth spelling) spread made from yeast. UK people can use Marmite instead.
Results (lower is better)
No compound at all 0.66°C/W.
White zinc oxide thermal compound 0.50°C/W.
Arctic Silver 3 thick layer 0.50°C/W.
Arctic Silver 3 thin layer 0.48°C/W
Arctic Alumina 0.50°C/W.
Cooler Master PTK-001 0.48°C/W
Cooler Master HTK-001 0.50°C/W
Nanotherm Blue after application 0.56°C/W.
Nanotherm Blue left overnight 0.51°C/W.
Toothpaste 0.47°C/W.
Vegemite 0.48°C/W
Plain water 0.41°C/W
So any compound at all is better than none, and the difference between expensive and cheap compounds is insignificant. He only used the joke items, Vegemite, toothpaste and water, to illustrate that almost anything between the cpu and cooler is going to aid heat transfer. Of course these substances are going to dry out pretty quickly and become useless or worse than useless.
He did note that if you are doing lots of cpu & heatsink tests and you are going to take the cooler off again in a few hours, you may as well use toothpaste to save money.
He also notes that if you think that choice of heatsink compound is going to make the difference between overheating and not overheating you must be running a pretty *censored* marginal system.
I was involved in electronic manufacturing for 20 years and his methods and results look good to me. I think that the market in heatsink compound is one of those "toys for guys" things, and also there is a good amount of superstition involved, similar to carrying a rabbits foot or 4-leaf clover to prevent bad luck.