For me, the ideal hit point or vitality system for tabletop roleplaying games involves the constant threat of engaging consequences while also mitigating the disastrous influence of luck. Using OD&D and sticking to the three little brown booklets comes close to this ideal when run in a certain manner, but still perhaps gives luck too much influence at first level and creates too much of a hit point buffer at mid to high level. Put another way, I want a system that encourages players to always care about combat consequences but rarely if ever shanks without warning. And, of course, the system must be fluent, easy to use, plugged in to the core flow of play, and require minimal bookkeeping.
Playing Kingdom Death gave me some ideas regarding ways to build a combat system that better prioritizes these goals, and that influence should be clear in the following sketch. As written, it may be too invasive to just trivially drop into a game using a B/X type engine, especially given that it requires replacing traditional armor class with ablative armor, but I think it would be possible.
Before anyone gets all up in my business about the dynamics of real armor or wounds, I want to emphasize that realism is a relatively low priority apart from maintaining predictable fictional consequences, necessary for allowing creative problem solving. Instead, the point is to create rules that facilitate choices and consequences while reinforcing the overall feel of the kind of survival fantasy that is my preferred mode for tabletop roleplaying games. This system assumes the turn structure of the Hazard System.
Mettle
Rather than hit points, player characters have mettle, which can be both bound to hit locations (all player characters have this kind of mettle) or floating (for tougher characters, those with high constitution). Instead of taking damage, player characters mark mettle boxes. The hit locations are head, body, abdomen, arms, and legs. Each location has two points of mettle except the head, which has one point. Additionally, player characters have a number of floating mettle points equal to the constitution modifier. These points can absorb damage to any hit location.
Defending
Currently, Hexagram uses active defending—blocking or dodging—sort of like this, rather than resolving monster attack rolls versus player character armor class. The details of the monster attack step are less central to the mettle and trauma system, and any method to decide if an opponent hits a player character in combat should slot in fine here. Even just leveraging the saving throw system seems like it would be a totally functional system for determining whether a character risks taking some damage, bringing armor, mettle, and so forth into play.
Armor
Player characters can equip pieces of armor to any hit location. Armor is ablative, meaning that it reduces incoming damage. The protection offered by a piece of armor maps roughly to the traditional light, medium, heavy (or leather, chain, plate) scale with light armor offering 1 point of protection, medium offering 2 points, and heavy offering 3. Since player characters have five hit location slots, they can mix and match, for example by wearing a heavy visored helmet but a light, boiled leather breastplate. Equipped armor still takes a gear slot, so piling on protection comes at the cost of lower versatility. Additionally, player characters act at a disadvantage when wearing armor with protection higher than the strength bonus.
Damage
When a player character takes damage, determine hit location randomly, subtract armor protection from damage taken (minimum zero), and then mark off one mettle slot for each point of damage remaining. I assume that the magnitude of damage is generally around 1d6 (following OD&D flat damage). If at any point a player character takes damage and has no remaining relevant mettle, then the character is in danger and must roll for peril. This is the step that can potentially lead to serious consequences, including character death.
Hit location (1d6): 1 head, 2 legs, 3 arms, 4 abdomen, 5-6 body
Peril (1d6): 1 messy death, 2-3 bleeding, 4-5 fracture, 6 sprain
Fractures disable the affected hit location. For example, a character with a fractured arm can no longer effectively wield a weapon using that arm. Healing fractures requires magic or taking a haven turn to recover (which would require retreating from a dungeon to town). Sprains work similarly but player characters can recover from a sprain by resting for a single dungeon turn (so, in effect, sprains only influence the current combat).
Other than being messy, gaining the first bleeding condition has no direct result. However, getting the bleeding result again, even to another hit location, means the character bleeds out and dies.
Trauma
Injuries are tricky to handle well in tabletop RPGs. On the one hand, they can make characters much less fun to play even for players on board with working out the implications of player character hardship. On the other hand, a fictional consequence is almost always more engaging than simple HP attrition through both adding narrative color—fan of blood—and changing the context as appropriate—a tiled floor slippery with blood. Further, while permanently changing settings and characters through play is satisfying, ruining characters is generally not. You can’t lose an arm in Dark Souls, and if you did I imagine the common response would be to restart the game. That would be a hardcore lose condition. For this reason, the peril table includes immediate fictional consequences beyond something like HP loss, such as heavy bleeding or broken bones, but defers the possibility of permanent disfigurement, the control of which falls to players through the grit system, described below.
Grit
Bleeding and fractures count as trauma, and surviving trauma strengthens tough characters. Player characters that survive a trauma can mark a grit box during recovery in a haven. Characters have a number of grit boxes equal to the constitution modifier. When a player marks a grit box, they should note how the trauma has permanently marked the character. This could be a scar or something else, and is entirely up to the player, but should be fictionally appropriate to the particular trauma (this would be a good place to insert fantasy prosthetics if such are setting-appropriate, such as necromantic grafts or enchanted wooden limbs). In effect, grit slots are like unlockable extra mettle slots.
False Machine has some creative ideas for scars here.
Implications
This system feels mechanically quite perilous. A mettle slot is roughly equivalent to one hit point, meaning that on average marking 1d6 mettle slots (expected value: 3.5) results in peril for all hit locations lacking armor. However, five-sixths (83%) of peril results are non-fatal initially. Strictly speaking, a one-shot kill is still possible, but is statistically much less common than the OD&D case of 1d6 damage versus 1d6 HP, and could easily be entirely eliminated if desired (such as by changing the messy death result to unconscious and dying, with final death occurring at the end of combat lacking miraculous intervention). Also, the odds improve dramatically with some armor while still maintaining the threat of real consequences.
In terms of complexity creep, this system requires an extra roll to determine hit location if an opponent hit is successful. So, there is a small increase in complexity, but the overhead seems minimal, which I have confirmed in preliminary play testing. The peril step replaces what I would otherwise run as a saving throw versus death, and that is an uncommon occurrence. Tracking the mettle and hit location slots does require a little help from the character sheet, but that seems manageable (see the character record sheet excerpt above).