Initiative Rules, and How to Pick a Fight 3

There’s a widespread acknowledgement, I believe, that the initiative rules found in D&D, and indeed in most games, are not great and even detract from the game, but they are the least-worst option available. In this post I’m going to talk a bit about the problems with the current model and some ideas of my own. At least, I think they’re new ideas to this conversation – there’s no knowing what has been discussed and discarded by other design teams.

The Current, Default Model

As it stands in D&D 5e, and essentially unchanged since 3.0, each person at the table rolls a d20, adds their Dexterity modifier and possibly other modifiers, and reports these values to the GM. The GM writes them all down – on paper, on index cards, on a dry-erase board, or in the group’s online-tabletop app – and the highest roll goes first. Initiative proceeds downward until you get to the last initiative value, then you start over again from the highest value. By default, you don’t reroll initiative each round, so you stay in that action order for the duration of the encounter.

5e plugs a few additional elements into this system.

  • Initiative Count 20 is a special position within the order. This 20 loses all ties, in case another player has rolled that value. On this initiative count, environmental effects such as lair actions are resolved. It’s important to know when such things happen, though any number would have been equally fair. Presumably, 20 was chosen so that quick PCs can possibly beat it, but even in a one-round takedown of a major boss (which should be rare), at least the lair actions still go off.
  • A very small number of possible characters care strongly about going first. Here I’m thinking of Assassin rogues, who rely on having a higher Dexterity modifier than most NPCs. They can give themselves a big boost in the odds with the Alert feat, if they want.
  • The duration of effects in 5e (as well as 4e) starts on your turn and usually ends either during the target’s turn (beginning or end), or during your next turn (beginning or end). Many, many spells and other effects are balanced on this timing. If you rerolled initiative in each turn, you could activate an effect “early” in one round and not act until “late” in the following round, greatly increasing its effectiveness. In the reverse case, you might completely eliminate it. Assuming these two outcomes will balance each other out and take care of themselves might work for some people, but to my mind it’s accepting a lot of unsatisfying outcomes.
  • Legendary actions refresh at the top of a legendary creature’s turn. They can spend actions only immediately following another creature’s turn. (Weird, it could rapidly dump its LA’s by interspersing them between its allies’ turns. I wonder if that is consciously intended. Moving on.) This creates a situation where, if you rerolled initiative in each turn, a legendary creature could act very late in one round and first in the following round, wasting that set of legendary actions.
  • It’s a Dexterity check, but there’s no way to gain proficiency in it. Still, bards can apply Jack of All Trades to it just fine. (If you already knew this, pat yourself on the back; I have found that this is not common knowledge even among serious 5e aficionados.)
  • The designers made the conscious decision to cut the Delay action from 5e, for a whole list of reasons that boil down to “shit or get off the pot.”

The system has its issues nonetheless.

  • The initiative roll interrupts the flow of the story, in what Adam Koebel has described as a smash cut – a phrasing I immediately adopted. The excitement and tension of blades flashing, arrows and spells flying, and people doing all that lovely bleeding is muted by a minute or two of bookkeeping. Not really ideal.
  • As a result of that smash cut, some things can happen that really mess with the flow of the narrative. What do you do when you’re in a tense standoff, exchanging banter or whatever, and suddenly the character who had already drawn down on the enemy just… lets fly? No one would act until that character does, because they were all fine with continuing the conversation (let’s assume for a moment that they’re also willing to fight now that it’s Stabs O’clock), but the character who triggers the fight rolls low on initiative. It doesn’t really make any sense to have the person who decided to make it a fight go anywhere but first, especially if the action required is as minimal as unclenching three fingers.
  • The absence of an option to rearrange initiative outside of the Ready action (which costs both your action and your reaction) does some specific, not-great things to attempts at teamwork and chaining effects. NPCs don’t have this problem, because (at least in 90% of the combats I’ve ever played or run) NPCs act on one collective initiative; breaking up the DM’s turn carries a lot of mental load without an interesting payoff. YMMV, sure.
  • It sure does privilege high Dexterity scores. This is sensible in the case of the Assassin rogue, but I’ve seen so many cases where the high-Dex skirmishers wind up just Readying or Dodging or anything else that gives the more Strength-based defender types a chance to move in. Or, when they don’t, being… a trifle put out to discover that they are the only targets available for the NPCs, and getting smeared. The issue here is really about quick characters not really aligning with the team’s tactical priorities, and the PCs having no valid solution other than a bunch of Readied actions or simply wasting those good rolls completely.

There are all kinds of solutions available: in the DMG, house rules on blogs and gaming sites (ahem, I’m going to add to that list), and even in other tabletop games. After all, only PBTA games get away with not having any initiative system at all, and that boils down to the GM’s sense of narrative timing shifting the camera to another player after each player has had a salvo of actions. A lot of the solutions found in other games are particularly interesting (Balsera Initiative, WHFRP 3e, and Shadow of the Demon Lord‘s fast/slow split come to mind), but require a deep rework of all of D&D to implement gracefully. Actually, WHFRP 3e’s system might be easy and smooth to patch in, in a limited way. Of course, with a human running the game, there can be a much higher tolerance for graceless implementation – then it’s just a tradeoff.

To that end, I’m proposing two alternate initiative approaches and two new actions.


The First Player Button

In many board games and card games, there’s a marker that passes clockwise around the table, signifying which player is first to act in the next round. To implement this in D&D, the players roll off to see where it starts at the beginning of each session; this is an unmodified d20 roll. The winner receives the First Player Button. In the first encounter of that session, that player acts first, and play proceeds clockwise. At the end of the encounter, the First Player Button moves one position clockwise at the table (or next in alphabetical order, or whatever makes sense for your arrangement). The First Player Button can go to the DM as it would any other player.

Obviously, this screws the Assassin. Therefore, rewrite the Assassinate feature as follows. “On your first turn in any encounter, choose a target to roll a contested Dexterity check, with a +5 modifier if you have the Alert feat. If you win this contested check, gain advantage on your attacks against that creature until the end of your turn.”

The Alert feat itself needs adjustment, since that +5 modifier mostly doesn’t do anything now (other than the thing I just said for Assassins). For this house rule, change it to “Increase your Dexterity or Wisdom score by 1, to a maximum of 20.”

Anything that would happen on initiative count 20 for environmental actions now takes place when the initiative passes the point in its cycle opposite the DM – that is, as far from the NPCs’ turns as possible. Because why not.

All of the class and subclass features that restore a class currency “when you roll initiative” are of course changed to “immediately before the First Player acts.” Any source of advantage on initiative rolls is dead in the water, because it’s tedious and contrary to the goals of this rule to give them two chances at the start-of-session roll-off. I don’t have an immediate solution, other than maybe an opening-round damage kicker.

If the party splits up and a group that does not include the First Player Button gets into a fight, play starts with the first player clockwise from the Button. The Button does not move at the end of the encounter, since the person with the Button did not act. If you have a bunch of encounters with these split groups, and that’s probably not a great idea unless you’re running Sense8 D&D, you should give each group their own First Player Button and treat them as fully independent teams. When they come back together, just do whatever seems reasonable and fair.

What I like about this model is that everyone gets a chance to go first at some point, rather than it mostly being a foregone conclusion; this pushes a little more variety into which players wind up defining the tactical battlefield with the earliest actions. It also eliminates the need for initiative rolls to intrude on the flow of the narrative. Instead, the DM just looks at whoever has the Button (after desperately trying to remember if it moved at the end of the last encounter) and asks, “What do you do?” or, if the DM has it, opens fire. It should slightly speed up play in every round, as the DM no longer keeps an initiative list to tell people when to act – they can just look around the table and know. It’s a little more explicitly gamist, but if you’re getting a simulationist warm-fuzzy off of “high Dex goes first,” then that’s kinda weird and I don’t share your gameplay priorities.

Obviously, it won’t work if your players are inclined to switch seats during the session. There are plenty of problems it doesn’t try to solve, as well.


One Initiative Roll Per Session

Roll initiative in the normal way once at the start of the session. That is the initiative order for the players for that entire session. Only the DM rolls new initiative in each encounter.

This sort of pushes Assassins into having good sessions or bad, which isn’t something you normally want to see; on the other hand, it’s only advantage against their first-turn target. It’s a loss, but not a crushing one. Having the DM roll fresh initiative for the NPCs in each encounter should help to make sure the Assassin still goes before them some of the time, if only on the strength of a dominating Dex bonus.

This doesn’t break the Alert feat or initiative count 20, and I think anyone can see that “when you roll initiative” becomes “when the GM rolls initiative.” On the other hand, any source of advantage to initiative (the Revised Ranger, for example) sort of works, but it’s not pretty – this probably amounts to less throughput than intended.

What I like about this approach is that it solves for the smash cut (the DM rolling one die in semi-secret is minimally invasive), while breaking absolutely as few things as possible, which is a good selling point for remembering how to apply the whole house rule. The DM (or whoever manages this at your table) still needs to record initiatives and tell people when it’s their turn, and high-Dex characters still trend heavily toward the top of the pack. By giving the DM new initiative rolls every encounter, it gives the DM more chances to go early in an encounter, diluting the player-side advantage of having the most chances to go early. This preserves some of the uncertainty of when the NPCs will kick into action, and thus how exactly to commit to the fight in the opening round.


New Action: Start Some Shit

Alternate name: Cry Havoc!

This action is for times when the DM hasn’t called for initiative because it’s not yet clear that there’s going to be a fight, but you’ve decided that there will in fact be a fight. See my commentary above for the difficulties in parsing this in most initiative systems.

When you are in a tense standoff and have a weapon in hand (with ammunition loaded, if necessary), or when you are part of a tense standoff but your opponents cannot perceive you, you can decide to start some shit (or cry “Havoc!”). Only one character in any encounter can do so; if there is a tie and it is not clear who has declared first, neither side succeeds and initiative proceeds normally.

When you start some shit, you immediately make one attack that is resolved before the DM determines initiative order. You cannot move prior to making this attack. However, if your opponents cannot perceive you, you can move up to your speed, and you can cast a spell as an alternative to making an attack. When the DM determines initiative order, apply a -10 penalty to your Dexterity check. If you are using an initiative system that does not involve a new initiative roll at the start of the encounter, the initiative order skips you in the first round, and you take your normal position in the initiative order thereafter. When an NPC starts some shit, that character skips their turn in the first round, and otherwise acts on NPC initiative.

(NB. Because you are in a tense standoff and your enemies are generally aware of your allies, starting some shit never results in surprise.)

As a variant of this rule, which carries certain tradeoffs in effect, use this instead: When you start some shit, everyone rolls initiative immediately, and you gain a +20 bonus to your initiative roll. (Don’t use this variant if your initiative system does not involve a new initiative roll at the start of the encounter.)


New Action: Hold a Hostage

One of the other issues that emerges from initiative order is that a knife to a hostage’s throat is a pretty trivial threat, undermining the tension of a lot of scenes. Ideally, the hostage-holder would Ready an Action and use their reaction to carry out their threat, which would have some reasonable chance (not guaranteed) chance to kill the target. By a strict parsing of the rules (not really recommended in this case, because the rules serve the narrative and not the other way round), this is impossible, since you can’t Ready an Action until you’re taking actions, so you would need to be in initiative order perpetually. Also, it’s difficult to deal enough damage with that readied action to make it the threat it should be; also, the whole image of the thing wants a knife or flintlock, not a greatsword or whatever. A long while back, I wrote the Subdue by Threat rule as one approach to this situation, but that’s more for subduing someone during a fight than holding a hostage during a tense standoff.

When you have a creature grappled and you are armed with a light melee or ranged weapon that deals piercing or slashing damage, but have not yet entered combat (because the grappled creature is not resisting you), you can use your reaction to any trigger you like to attempt to immediately kill the grappled creature. Make an attack against the creature with advantage; ranged attacks do not suffer disadvantage for this attack, and thus gain advantage normally. If this attack hits, it is automatically a critical hit, and you add your level or (for NPCs) twice your CR to the damage dealt.

As a reminder, you cannot use this reaction if you are denied reactions, such as when you are surprised, or when the action that would trigger your attack incapacitates you (for example, when you are killed in a single attack). You also cannot use this reaction if the action that triggers it ends any of the qualifying conditions to hold a hostage, such as ending your grapple or causing you to no longer be armed with a light melee or ranged weapon that deals piercing or slashing damage. Once you complete this reaction, initiative order proceeds normally.

PCs want to be heroes, and a majority of PCs (in my experience) specifically want to be Malcolm Reynolds, so don’t expect this to stop PCs from attacking someone holding a hostage. The goal, instead, is to establish the actions that break the hold, and if any of those fail, to give the threat some decent teeth. If you play with this for a bit but discover that it’s not getting the job done, try increasing the level-scaling damage kicker to twice your level, and thrice the NPC’s CR, or higher. This action is intended for use alongside Start Some Shit, since they handle both sides of a situation where initiative order does a poor job of serving the narrative, but you want initiative order to kick in soon once it becomes a “normal” combat.



The One Roll Per Session model is a more-involved version of the DMG’s Fixed Initiative variant. What it offers over Fixed Initiative is that there is session-by-session variation, that advantage on initiative checks does something, and that it’s much more likely that someone beats initiative count 20. At least in theory, I like leaving in the encounter-by-encounter variation in when the NPCs act. It’s also not 100% clear to me how one handles NPC initiative in this variant, especially if Dex scores vary within the NPC group. (That’s a general problem with lumping all NPCs together, though.)

Some of you may be wondering, “Why do there need to be hard rules around Starting Some Shit or Holding a Hostage? The DM’s ruling on the spot is good enough for me.” To that I would say, a game situation that involves maximum in-character tension is the worst time for any kind of rules dispute, as the DM’s ruling becomes a matter of whose creative agenda wins out. It’s the kind of situation where the DM wants to appear strictly neutral, a moderator of the forces in the fiction rather than the mastermind of the opposition – despite being both simultaneously. That’s when it’s best to get all of the rules out there ahead of time, though “ahead of time” can really be as late as “right before somebody starts some shit.”

I’m interested to hear about unforeseen consequences or feature interactions that I’ve overlooked for any of these initiative rules or new actions. I absolutely might have missed something, even something big.

Leave a comment

Your email address will not be published. Required fields are marked *

3 thoughts on “Initiative Rules, and How to Pick a Fight

  • Will

    I think I might try the second option tomorrow night, and allow players to reroll 1s after taking their surprise penalty round.

    Was contemplating trying the Star Wars ffg model – initiative count as player slots, with the party choosing who takes what, rather than strict pc1 goes at count 2 etc. Wonder if that might allow for a happy medium combined with the one fixed roll method as no one pc would be screwed over all the time.

    • Brandes Stoddard Post author

      There’s nothing wrong with the FFG model, though I don’t recall if you reroll initiative each round or if it carries forward through the whole encounter. If you reroll, it will make some odd things happen with spell duration that may be undesirable. The main reason they bug ME, of course, is that they don’t apply equally to the NPCs (who all act in one big cluster).

  • 1Mac

    1.) It is not obvious to me that the character Crying Havoc should always get to act before initiative or have any sort of automatic advantage at the start of combat. It seems to me that the way to think about this setup is as a sort of social Stealth check, using Deception instead of Stealth. I get that 5e’s Stealth rules are a mess, but the point is that just as a clanking paladin isn’t going to sneak up on anyone, an oafish thug isn’t going to catch anyone off guard. They are setups that not all characters can reliably take advantage of, so some sort of check seems in order here.

    2.) Have you ever checked out the “tick” system Hackmaster uses in place of initiative? As presented it’s a little fiddly for my tastes, what with every weapon having its own speed, but the basic idea seems pretty slick. I think a simplified version could fit DnD pretty well.