As far as traps go, my feeling is to keep traps themselves relatively simple. Elaborate traps tend to be better modeled as skill challenges with the core trap mechanism built into the consequences/rewards of the challenge.
That is I would consider a pit trap to be a trap, pure and simple. But if the trap involves multiple puzzle-like elements then it is probably best handled as a skill challenge where different skill checks will provide the information needed to thwart it. Usually this type of trap doesn't have just one way to get around it, nor do they usually HAVE to be straightforwardly disarmed (though some can be). That brings up another advantage of using the skill challenge mechanism for this type of thing, if you treat a trap as A trap, then logical a rogue should be able to entirely disarm it in one roll with Thievery, which may work out as being a bit cheap when you put a lot of hard work into designing it.
Always remember too that you can quite effectively combine skill challenges with monsters. There are examples in the DMG, but one I currently have slated to add to a dungeon the party will soon enter is a combination of rock climbing, repelling attacks from flying creatures, and possibly some other elements I haven't decided on yet. That should provide some interesting challenge. I think the climbing part will be worked out such that the party will most likely be climbing down an unstable slope. They can just plain climb, or rappel, use pitons, rope themselves together, etc. The worst case scenario for them is plunging into the depths. If they are smart and set up anchor points and whatnot then if all goes well they can avoid that. Dungeoneering checks can give them the ability to set pitons in good locations, Perception checks can let them see where it is a bad idea to climb, Athletics checks can be used to jump from rock to rock, etc. Enough success rolls means the climb is successful. Failures will of course be bad, but even then I'll provide some ways to get saves or acrobatics/athletics checks to allow individual characters to avoid the big fall. I think how I will work it is to provide 'graduated failure' based on the number of successes before the party fails. So if they fail on the last check, then maybe they just loose some of their equipment. If they are short a larger number of success rolls then the disaster strikes higher up the cliff and the fall will hurt more. The monsters attack will probably not be a HUGE factor, but more adding to the tension and possibly knocking a character off if he/she is attacked at a bad moment, with maybe a single artillery monster at the bottom of the cliff that takes potshots at everyone as well, giving them a real urge to go quickly so they don't get whittled down too much before they can take it out.
Sometimes in more complex situations you can make a kind of a flow chart of how the challenge works. As the party progresses through the challenge they will arrive at different choice points. Each choice point will have potentially several branches, one for each skill applicable at that stage of the challenge. Then you can have a branch for each that depends on failure/success, or the result of a roll simply changes some variable like giving a bonus to another skill check, etc.
I also like 'partial success' as a principle in challenges. For instance a negotiation with the theives guild could have a minimal threshold for success, but instead of a fixed number of failure rolls just add to the price they have to pay to bribe the theives for each failure. If the cost exceeds what they can pay, then the whole thing breaks down and warfare can erupt!
