I think there is an alternative that is being implicitly disregarded here. Character death is indeed extremely disrupting in most RPGs and one way to deal with it is avoiding it altogether. The alternative course of action is, of course, trying to eliminate some or all of the disruptiveness of character death.
So, let's talk about immortality.
I am aware of at least one computer game that experiments with this: Planescape Torment. To quote the vision statement:
The player can have horrible, flesh-rending things happen to his character
on-screen – even die -- and he will get better. What happens to monsters above
can happen to the player – except that the player will regenerate. The player‟s arm
can be loped off, his legs severed, his skin flayed from his body, and he‟ll be fine
in a few hours (and perhaps a little stitching by the Dustmen). In some instances,
the player will need to lop off his own extremities for use as weapons (arms make
great clubs!) or free himself from manacles – he can use his deathless state to help
him solve problems during the adventure.
The designers of PS:T started off using immortality to get around the problem of character death (which is similar but not quite the same in videogames as it is in RPGs) and ended up basing the whole game premise on that. Note that in PS:T the cost of character death is almost zero: you just have to walk from the mortuary to the place you died. Also note that this doesn't create any complications if some but not all of your characters die.
Making it so the characters rapidly regenerate after their apparent death is not the only way to alleviate the disruptiveness of character death: Another option is a timeskip. Imagine a main plot that spans centuries and characters that cannot die of old age. When the characters are killed, they are resurrected (or reincarnated) some time (months, years, decades) later. Note that the main plot needs to remain unaffected of these deaths; for example a conspiracy the magnitude of which could be revealed bit by bit, at any time. In this case, the cost of character death would be sentimental (stakes are lost in their absence), narrative (subplots are not fully revealed), or even mechanical (equipment is lost or becomes obsolete). What if some but not all of the characters die? You can deal with this by not allowing it to happen in the first place, by designing challenges so that the skills of all the characters are needed to survive. If the worst comes to the worst, you could provide incentives for the survivors to wait for the dead characters to return.
A third option is to say that death is a journey. You can think of characters as immortal souls so, when they die in one life, they can carry on in the next one. Maybe they'll try to return to their birthworld, you could work with that. Or they could die (again) trying, falling deeper into the realms of death. That would more appropriate in a horror campaign, of course. Or, finally, they could come to terms with it and try to live their new lives. In this case, the cost of death is either all their old lives or the time they take to return and the stakes that are lost in their absence. What if some of the players survive death's clutches? One solution is to show them that death is not the end and give them an opportunity to follow their friends without losing their dignity (for example by crossing a magical gate). Still, I'm not going to lie: this may turn sour.
Finally, immortality is not the only way to make character death more manageable. You could replicate the save/load mechanism that exists in videogames, for instance. Imagine a spell which allows you to set a savepoint in time. Then, if one of the spell's recepients dies, the spell is triggered and the characters are transferred back in time. This one is particularly advantageous, since you can set the cost of dying to whatever you want by specifying the spell's casting requirements and costs. This also circumvents the problem of having some players die and others surviving, since the spell affects them all.
In conclusion. Make no mistake, this isn't something that should be integrated in every game you play. Lowering the cost of death is very visible to the players. Do it often enough and they'll think you're obsessed with the theme of immortality or something. However, if done sparingly, it would solve the problem of the disruptiveness of character death without feeling contrived. The important point is that it's easier for the players to buy in since it's part of the premise of the game and not something you pull out of a hat when you need it.
Post Scriptum: What's so wrong with avoiding character death?
Let me state right away that I don't think that it's all wrong to avoid killing the characters. But I do think that it creates a number of problems, so you, as a GM, need to be aware of them.
Peter Leppert's answer (and I apologise in advance for somewhat de-constructing it) gives you two ways out:
- The first is to set the stakes to something other than the characters' lives. This works well. But you can only rely on it most of the time. Once in a while, you have got to remind the players that being an adventurer is a risky career choice and that their characters are brave for doing it. Daring death is exciting (otherwise it wouldn't feature so prominently as a plot device or a game mechanic) and it would be a pity to just forget about it.
- The second way is to pretend that that the character's lives are at stake, only it turns out they're not really. Sure, the first time your characters don't retreat in a hopeless battle you could have a set of NPC heroes save them and embarrass them (so make them think their lives are at stake but then change your mind and say it was their reputation at stake). But what would you do the second time? Or the third? It could end up feeling contrived. The fact of the matter is, you just can't come up with good excuses to pull the ol' switcheroo often enough.