CRs and ELs have always been, at least to me, a bit nebulous. They're estimates based on the following:
1) That a CR/EL equal to party's level (or average or effective level, I'll define those terms later) will consume 1/4 of the party's resources.
2) Said party consists of four memebers.
3) Said party covers the four basic food groups of adventuring: a Cleric, a Fighter, a Rogue, and a Wizard.
4) Said party is likely to be prepared for the encounter in question.
That's quite a bit of assumin'.
Now, average level should be obvious (add up the levels, divide by the number of party members). For effective level, I figure out the party's encounter level (as per the DMG) and subtract 4. For a four PC party of equal level, the average party level (APL) and effective party level (EPL) are the same. The EPL changes if the party doesn't fit this model.
For instance, take a group of 5 PCs, all 6th level. The APL works out to be 6, but the EPL is more like 7. This tells me that I can use an EL 7 encounter as a fair challenge, although a single CR 7 foe might be a bit tough. Add a 7th level PC to the same group, and while the APL rounds down to 6, the EPL is equal to 8. Here's where things start getting tricky, and I have to look past the stats to plan out an encounter.
Not all CRs or ELs are equal. Let's take the basic party of four as listed in item #3 above, set them at 4th level and put them in two separate encounters. The first is a Minotaur (CR 4). The second is with eight Orcs (CR 1/2). Both are EL 4, but both fights will have different results.
Facing a single opponent stacks things to some degree in favor of the PCs. In the case of the Minotaur, the Fighter closes while the rogue sneaks or tumbles into position fo a sneak attack. The Wizard blasts while the Cleric heals or smacks as necessary. It'll might go more easily than the fight against the Orcs, since the PCs will have to split their efforts.
As an example, the game I'm running consist of five 2nd level characters (Bbn, Clr, Ftr, Rgr, and a Wiz). They found a smuggler's cove the hard way, and the ensuing brouhaha turned out to be two encounters. The first consisted of most of the smugglers, seven War 2 (Cr 1 ea, EL 6), and later a group consisting of one smuggler (War 2, CR 1), two Gnolls (CR 1 each) and a 4th level sorcerer (CR 4, for a combined EL of 6 as well). Just based on these numbers, there should have been PC death in the first encounter. Instead the PCs survived both, albeit very spent (which goes without saying).
This goes into what else influences the encounter beyond the numbers -- tactics and the environment, or rather, who controls the good ground. In the first encounter, the PCs bottlenecked the enemy on a stairwell, with the Fighter and Barbarian holding ground near the base and the Cleric right behind them, healing at the ready. The Ranger(who has Precise Shot) and Wizard(who has a reputation for being a hell of a shot thanks to hot dice) held the high ground and could bring missile fire onto the smugglers. Five NPCs were slain, and the other two fled.
The second fight went in favor of the PCs due to a tactical error and an element of luck. The sorcerer took the time to cast Invisibility on himself, the two Gnolls and the remaining smuggler (who was not in the bloody melee because he stopped to grab some spears to throw). They hid themselves at an intersection, waiting for the right opportunity to either strike or flee. When the Fighter and the Barbarian took point, the Sorcerer decided to strike and blasted the two with a Color Spray. Unfortunately, the DM (me) didn't plan much after that, assuming (much like the spell caster) that both of them would fail their save. The Barbarian failed, the Fighter didn't. One five foot step later, the Fighter drops the sorcerer in one strike before the Gnolls could intercede. Things went downhill for the bad guys after that.
Of course, shortly after that the party ran into half a dozen skeletons, and got real worried until the Cleric (also a halfling) walked up and nuked them with that holy thing they do so well.
For another (higher level) example, a party of four 19th level characters (Cleric, Fighter, Wizard, and Ranger), fresh for a fight, suffered three deaths in a fight with a CR 20 Dracolich. The only reason the Wizard survived was because he ran like hell and could fly faster than the creature.
While it was, by the numbers, a challenging encounter against a single foe, the party, freshly rested and fully healed, should have been able to deal with the threat. So what happened? Multiple things, really...
The first was lack of preparation. The party had been encountering nothing but undead in the Tower of the Dragonlord while looking for the Gold Orb of Dragonkind. After encountering two Red Dragon skeletons, you would think that something worse than that would be in store. But instead of focusing their abilities on slaying undead, they kept with a more versatile complement.
Then there was bad tactics. The Dracolich had plenty of room to maneuver and fly. The characters had magical flight, which evened that out, but the Cleric, the Fighter and the Ranger all closed for melee in one solid rush. When I started asking for Fort saves as the dracolich hit (for the paralytic touch), that should have struck the PCs as significant, seeing as they've faced a lich before and have high enough Knowledge skills amongst them to know what they can do. They could have retreated, and the thought did cross the Cleric's mind, but they didn't.
But what really did the PCs in was bad luck. Two natural 1s and natural 2 on the Fort saves neutralized the three PCs in close combat. Then came the coup de grace, since the Dracolich was determined to keep the Orb out of anyone's hands.
So as you can see, while the CR/EL numbers can give a good thumbnail of how difficult an encounter may be, there's more to it than just the creature stats. Terrain, tactics, preparation and even psychology all make a difference. The key to designing fun and challenging encounters (or misleading ones) is to consider the context of the whole.