以下引用暴雪蓝贴：
http://blue.mmo-champion.com/28/16473618356-116-armor-pen-only-giving-66.html时间2009-4-18
Okay, here is a fairly technical explanation we put together for how armor pen works.
We didn’t want Armor Penetration Rating to be too powerful against low armor targets, like it had been in BC. We also didn’t want Armor Penetration Rating to be too powerful against high armor targets.
So, we decided on a system where there is a cap on how much armor the Armor Penetration Rating can be applied to. So, the first X armor on the target is reduced by the percentage listed in the Armor Penetration Rating tooltip, and all armor past that X is unaffected. Another way of understanding that is we multiply the percentage in the tooltip times the minimum of the two values: the cap, and the amount of armor on the target after all other modifiers.
Computing the cap is a little tricky unless you are already familiar with how World of Warcraft armor works. There is an armor constant we’ll call C. C is derived as follows (in some pseudocode):
If (level<60)
C=400+85*targetlevel
Else
C=400+85*targetlevel+4.5*85*(targetlevel-59);
For a level 80 target, C=15232.5. For a level 83, C=16635.
The cap for Armor Penetration then is: (armor + C)/3.
A level 80 warrior creature has 9729 armor. C=15232.5. So, the cap is (9729+15232.5)/3=8320.5. Let’s say a player has 30% armor penetration from armor penetration rating and no other modifiers that complicate the calculation (talents, Battle Stance, Sunder Armor, etc.). The game chooses the minimum of 8320.5 and 9729, so 8320.5. That is multiplied by 30% = 2496.15, and so that much armor is ignored. The effective armor on the target is 7232.85 (9729-2496.15). From a player point of view, the armor penetration rating didn’t ignore the full 30%, but instead ignored 25.66%. (85.5% as effective as expected).
These equations should help you be able to test and verify that Armor Penetration Rating is working correctly and as we designed. The tooltip is not actually inaccurate, as it states: “Enemy armor reduced by up to 30.00%.” That "up to" is key.
Please be sure to test without any other effects which modify the armor calculation (Battle Stance, Sunder Armor, Mace Specialization, etc.) as they may involve other systems that add additional complexity to the calculation.
设护甲常数C
如果目标等级小于60级
C=400+85*目标等级
如果目标等级大于等于60级
C=400+85*目标等级+4.5*85*(目标等级-59)
在WLK中，大多数情况C取以下两个值：
对于等级80的目标，C=15232.5
对于等级83的目标(Raid Boss)，C=16635
则，护甲穿透[上限值]=(目标护甲值 + C)/3
设有一位等级80的小怪，其护甲值为9729。
由上式，C=15232.5。因此，护甲穿透上限为(9729+15232.5)/3=8320.5。
再设一位玩家拥有30%护甲穿透(由护甲穿透等级算出，即人物面板显示值)，并且，没有其他影响计算的因素(如天赋，战士的战斗姿态，目标身上的破甲技能等等)。
实际在游戏中，8320.5和9729这两个值中的较小值会被作为目标参与计算的护甲值。
于是，目标实际被忽视的护甲值为8320.5*30%=2496.15。该玩家在攻击该怪物时，其有效护甲为9729-2496.15=7232.85。
从而，这位玩家会觉得，他的护甲穿透并没有达到标称中的30%之多，而只有2496.15/9729=25.66%。
以上公式有助于帮助你测试及验证护甲穿透系统是否像我们当初设计的一样正确工作。另外，面板提示并非不准确，因为提示说的是“敌人护甲降低最多30%”，其中“最多”两个字是理解这个问题的关键。
我英文不好，只是照搬某NXA网站的翻译。
http://bbs.duowan.com/thread-15387914-1-1.html