As stated before, the difficulty in finding an adequate power supply for your system is a direct result of the emphasis placed on often superficial numbers that can be manipulated in several ways by manufacturers.

The most commonly used number when advertising power supplies is the Maximum Power output. Sadly, most consumers in the market for a power supply base their purchases off this number alone. There are so many problems with doing this that I am not even going to attempt to name them all, but I will point out the most prevalent ones.

First of all, a quick physics lesson:
The majority of numbers you see on a power supply label are followed by one of three letters; W, V, or A.

W = Watts. A watt is a measurement of power.
V = Voltage. Voltage is a measurement of the difference in Electric potential energy between two points in a wire (electromotive force).
A = Amperes. An amp is a measurement of current.

To relate them all metaphorically: voltage is the amount of water pressure in a hose, amperes is the rate of flow in the hose, and watts is the amount of work being done on whatever you're spraying it at, per unit of time. They can be expressed in this equation (which is always true for DC currents):

V x A = W

Knowing this, you would think that due to the fixed voltage rails, strong currents on each rail will always mean a higher power rating and therefore a good product. This is true only to a certain extent. In the past, CPUs, video cards, hard drives, and other devices relied primarily on the 5V rail to deliver the high current they needed. This is why on older power supplies you see high amperage on the 5V rail, often times as much as 50A. However, modern PCs have swapped out the 5V rail in favor of the 12 and 3.3V rails to supply power to the high current components of a computer. Therefore, if you are using or purchasing an older power supply you will see a high power output, but not necessarily in the area you need it.

To illustrate the next reason why Maximum Power ratings are misleading, I have taken a picture of the label on my power supply to use as an example.

Now do the math by plugging those numbers into the above equation, and you will see that the "Total output wattage" value of 250W is not the 346W that you get by using the equation. This is the result of "cross regulation", where the combined power output of the 3.3V and 5V rails varies with the output of the 12V rail. Basically it means that each rail draws a different amount of power from the total output at different times. So when your video card and CPU are getting stressed when running a benchmark like 3DMark05, the 12V rail is drawing a lot more power, which means that the 3.3V and 5V rails are using less current than they would. Also note that although my power supply is rated at 250W, it is running my "recommended 300W minimum" ATi X800XT just fine, even when the card is overclocked.

A third and probably the most important reason is that different manufacturers make different quality products. You cannot tell whether or not a power supply is a good one just by looking at it, regardless of what it says on the label. One popular method of determining the quality of a power supply is its weight, which actually has virtually no correlation with performance. In a power supply the majority of the weight is coming from the transformer and the inductors, which physically are just long peices of wire wrapped in a coil. The manufacturer of the power supply chooses which components to use, and as such the quality of your power supply is very much at the discrimination of the company you buy it from. Suppose a manufacturer chooses to use a MOSFET that costs slightly less than a more expensive, yet smaller, lighter, and more efficient one. Not only does this mean the MOSFET weighs more, but it is also producing more heat (less efficient), which means that the heat sink attached to it is going to have to be bigger and bulkier to transfer the excess wattage lost as heat. To illustrate this point, take these two 550W power supplies:

1. Powmax PSAG550A
2. Fortron FSP550-60PLN

If you think for one second that I would put that POWMAX 550W in my system, you are dreaming. Although it costs more than 5 times as much, the Fortron supply is not only from a reputable manufacturer, but also offers dual +12V rails, the advantage of which will be discussed later. While the POWMAX might last you a month or two before frying, you could get several years of use out of the Fortron, and probably not damage any of your other components in the process. This is no attack on the POWMAX company, they make some fine higher-end models, but the quoted one is not up to this standard.

Furthermore, most of the testing done by power supply manufacturers is at low temperatures, often times as low as 25° C, this lowers resistance, which increases with temperature, and produces inflated performance ratings. Not only does such a power supply label mislead consumers about its actual performance, chances are that it will also fail with increased operating temperatures. Another way manufacturers deceive customers through labels is the VAC Tolerances. Poor power quality is a symptom plaguing many circuits these days. Power ran through too many electric devices starts coming in harsh bursts, resulting in large instances of instability. As I sit here typing this, the lights in my room dim momentarily as the air conditioning unit in my house switches itself on. Unfortunately we have relatively little control over power quality, even subtle differences caused by electromagnetic interference can wreak havoc on the integrity of the power available at your wall socket. Manufacturers realize this and often rate their power supplies within certain tolerances. We also see a trend here among reputable manufacturers as opposed to the lower quality ones. Supplies from companies such as PC P&C can operate through much higher range of voltages and frequencies than certain other companies. It all has to do with the input regulation stage explained above, but we can combat these power fluctuations using a combination of surge protectors and Uninterruptible Power Supplies.

One last thing I would like to mention before moving on is what is called the Power Factor and Power Factor Correction. PF and PFC are figures you may or may not see when shopping for a power supply. A power factor is basically the ratio of how much power a device is actually using to how much power the device is supposed to be using. This ratio varies on a scale from 0 to 1.0, 1.0 being the best. Most power supplies have PF ratings of around .6 to .7. Adding "passive" PFC, usually either a capacitor (for inductive loads) or a an inductor (for capacitive loads), can increase a supply's power factor by roughly one and a half tenths. Adding an "active" PFC, such as a secondary switching circuit to condition the input power, has the potential to raise a power factor to almost perfect. This does not mean anything in terms of efficiency however, because the power is still being used and wasted by something in one way or another.