As mete said, when you raise the temperature of the blade form below the Mf point (approx. 200 for most simple steels, but as much as below -100F for complex steels), to between the Mf and the Ms point, some of the retained Austenite can convert to Martensite. This doesn't happen on the up swing , but when the metal cools back down from the temper (falling back through the Mf point) . That means that the new Martensite is untempered. A second temper (or third) will be required to finally have the blade get as much tempered martensitc structure as possible. More than 3 tempers is a case of diminishing returns in any steel we will deal with.
The more alloy ingredients, the harder it is to get all the Austenite to convert. With a simple steel ,like 1080/1095, all the austenite usually converts with the quench (hopefully to Martensite), thus a single temper will give you a tempered martensite blade.A double temper is a good rule to follow, though.
One thing I regularly notice is newer smiths trying to temper at a very low point. This is counter productive to what the temper does.The blade will not suddenly become soft as lead at 350F (or some imagined point). It has to go well above the Ms point to start getting a rapid decrease in hardness.A look at the HvsTT charts will show you that most carbon steels would be fine in excess of 500F. 1080, and O-1 both will be Rc60 at 500F. What the blade does gain when you temper at the right temp, is an increase in toughness.
Stacy