A problem that I have in GMing RPGs, and I imagine a lot of other people have, is reluctance to kill characters that players have become strongly attached to. I'll describe my evolution in how I have dealt with this.
The Law of Survivability
In its early days Dungeons & Dragons was intended to be played with a single character per person, with hirelings to beef the party up to a reasonable number. That number was closer to eight than to the four we saw in D&D third edition. Those numbers make a big difference, as "Lew's Law of Survivability" is intended to illustrate:
Note: this is about survivability of the entire party, but that should enhance individual survivability. The numbers are relative, that is, a party of 3 (3 squared = 9) is one quarter as survivable as a party of 6 (squared = 36).
The Law of Single Characters
And if a player has only one character, the GM is much less likely to allow that character to die, as I indicate in "Lew's Law of Single Characters”:
Hirelings earned half experience, and if the principal character got killed it was usually possible for the player to become one of the hirelings.
In those early days we didn't make up detailed backgrounds for characters. Sometimes they didn't even have names to begin with, as we let what happened in the first several adventures define the character and suggest a name. We were quite game oriented and not nearly so much story oriented.
I was the original GM in our group, but I wanted to play as well as GM, so I encouraged other players to learn to GM. This led in a large group to players using characters in the campaigns of several GMs at the same time. Hirelings per se were entirely dispensed with. Sometimes a player played two characters when there were not enough players to make a party of at least six and as many as eight. In some cases the players who regularly gamemastered got to play a second character while those who did not GM played just one—gamemaster privilege. There was always a “overall GM” who was in charge in case screwy things happened (which usually involved one GM giving too much "stuff" away).
Today we find many players who are much more interested in story than game, and who want to make a mark on the campaign with the story of their character. This frequently means that the player devises (usually with approval from the GM) an elaborate backstory for their character. I have never done this because it slows down the initial games, and I prefer to get people playing the game rather than worrying about the individual non-game details of their character, especially when there's a significant chance that the newbie characters will be killed with little hope of resurrection.
The Law of Character Generation
I've also seen that the more time a player puts into a character, the more incentive there is for the GM to keep that character alive. Hence "Lew's Law of Character Generation":
The Law of Imposed Stories
A strong corollary to the GM telling the players a story is that hardly any character will ever be killed - unless in service of the story. Hence "Lew's Law of Imposed Stories”:
To go back to the original point, my solution is to get characters into a game as fast as possible—which also seems to be the strategy in D&D fifth edition—while allowing players more than one character when that's appropriate. When players have several characters gradually progressing there are two benefits: it takes much longer (in real time) for players to reach higher levels, and if a player's character dies that player has several others to play, and he/she is not devastated the way they would have been with just one character.
I know that with the experience we have among the readers there have been other solutions to this, and I hope you'll reveal how you’ve coped with the "problem" of character death.[/excerpt][/excerpt]
The Law of Survivability
In its early days Dungeons & Dragons was intended to be played with a single character per person, with hirelings to beef the party up to a reasonable number. That number was closer to eight than to the four we saw in D&D third edition. Those numbers make a big difference, as "Lew's Law of Survivability" is intended to illustrate:
The survivability of a party varies with the square of the number of characters in it.
Note: this is about survivability of the entire party, but that should enhance individual survivability. The numbers are relative, that is, a party of 3 (3 squared = 9) is one quarter as survivable as a party of 6 (squared = 36).
The Law of Single Characters
And if a player has only one character, the GM is much less likely to allow that character to die, as I indicate in "Lew's Law of Single Characters”:
The more a player focuses on just one character, the harder it is for the GM to have that character die."
Hirelings earned half experience, and if the principal character got killed it was usually possible for the player to become one of the hirelings.
In those early days we didn't make up detailed backgrounds for characters. Sometimes they didn't even have names to begin with, as we let what happened in the first several adventures define the character and suggest a name. We were quite game oriented and not nearly so much story oriented.
I was the original GM in our group, but I wanted to play as well as GM, so I encouraged other players to learn to GM. This led in a large group to players using characters in the campaigns of several GMs at the same time. Hirelings per se were entirely dispensed with. Sometimes a player played two characters when there were not enough players to make a party of at least six and as many as eight. In some cases the players who regularly gamemastered got to play a second character while those who did not GM played just one—gamemaster privilege. There was always a “overall GM” who was in charge in case screwy things happened (which usually involved one GM giving too much "stuff" away).
Today we find many players who are much more interested in story than game, and who want to make a mark on the campaign with the story of their character. This frequently means that the player devises (usually with approval from the GM) an elaborate backstory for their character. I have never done this because it slows down the initial games, and I prefer to get people playing the game rather than worrying about the individual non-game details of their character, especially when there's a significant chance that the newbie characters will be killed with little hope of resurrection.
The Law of Character Generation
I've also seen that the more time a player puts into a character, the more incentive there is for the GM to keep that character alive. Hence "Lew's Law of Character Generation":
The longer players take to generate characters, the less likely those characters are to die.
The Law of Imposed Stories
A strong corollary to the GM telling the players a story is that hardly any character will ever be killed - unless in service of the story. Hence "Lew's Law of Imposed Stories”:
The more a GM treats their RPG campaign as storytelling session, the less likely it is that a character will die.
To go back to the original point, my solution is to get characters into a game as fast as possible—which also seems to be the strategy in D&D fifth edition—while allowing players more than one character when that's appropriate. When players have several characters gradually progressing there are two benefits: it takes much longer (in real time) for players to reach higher levels, and if a player's character dies that player has several others to play, and he/she is not devastated the way they would have been with just one character.
I know that with the experience we have among the readers there have been other solutions to this, and I hope you'll reveal how you’ve coped with the "problem" of character death.[/excerpt][/excerpt]
My players weren't too fond of henchmen because of the experience loss. The problem was finding the sweet spot between party size and experience gain. Sometimes that depended on the classes the PCs played. A dungeon heavy in undead favored Clerics back in the day thanks to "turn undead". A dungeon with areas wider than the typical 10' wide corridor needed more fighters, etc. 
