My post–high school Advanced Dungeons and Dragons group had a running joke—OK, we had about 600 running jokes, but one of them was that for any given encounter situation, there were always a plan A and a plan B. Plan A was “Get ’em!” Plan B was “Run!”

Fifth-edition D&D, with its inclusion of opportunity attacks, has made it curiously challenging to execute plan B.

This isn’t a brand-new concept. It existed in D&D version 3.5 and fourth edition, and many other tactical games, both tabletop and computer, incorporate opportunity attacks. But because of the turn-based nature of these games, a combatant who wants to retreat is confronted with a difficult and unpleasant choice: If the combatant uses his or her action to Disengage, then uses his or her full movement speed to retreat, the opponent can use its full movement speed to close the distance again, then use its action to Attack. But if the combatant uses his or her action to Dash, he or she risks getting struck by an opportunity attack upon leaving the opponent’s zone of control.

It seems unfair, and yet, as I learned from this Reddit post, retreating from combat is supposed to be difficult. It’s difficult in real life. Here’s the part that leaped out at me (emphasis in original):

Being able to disengage and retreat in good order was the sign of a professional, disciplined, and loyal army because it would invariably cause casualties as men were cut off, left behind, parts of the line losing coherency, etc. It also involved many men receiving and adhering to orders in the heat of battle.

In short, it was incredibly hard and it was the sign of a robust army.

Rearguard actions were common. This is where ‘blocker’ forces are used to essentially buy the rest of the retreating army time to fall back and regroup. Sometimes these rearguard actions were essentially suicidal in nature, where the assumption was those fighting in the delaying action were doomed, but not always.

There were many examples of successful rearguard actions that didn’t involve the delaying force to sacrifice their own lives, sometimes even turning the tables on their pursuers.

But having a rearguard involves manpower, willing soldiers, and supreme discipline/training on the part of those staying behind to fight, knowing they are the last ones to leave.

In many cases, the pursuers are just as exhausted as the retreating forces and following at a quick pace is strategically unwise or just plain impossible.

In short, even when executing an orderly retreat, combatants simply had to hope that their opponents wouldn’t pursue them, because pursuit could very well mean death for those who ran interference . . . one of many reasons why one is always advised to choose one’s battles wisely.

Let’s take a look at the three ways D&D 5E provides of getting out of a sticky situation: Dodge, Dash and Disengage.


The Dodge action imposes disadvantage on all incoming attacks from visible opponents (and also grants advantage on Dexterity saving throws). After Dodging, you have your full movement speed available; if you move out of your opponent’s reach, however, you’re subject to an opportunity attack.


The Dash action effectively allows you to double your speed by granting you your full movement speed as an action, then granting it again as your movement. If you move out of your opponent’s reach, you’re subject to an opportunity attack.


The Disengage action lets you move out of your opponent’s reach in an orderly fashion that denies opportunity attacks. After Disengaging, you have your full movement speed available.

Ludovic vs. the Bugbear

Ludovic is a level 3 fighter engaged in melee combat with a bugbear. He’s decided he doesn’t like his chances anymore, and he’s looking to get out of the situation. His armor class is 15. He has a normal 30-foot movement speed. So does the bugbear. The bugbear has +4 to hit, so it has a 50 percent chance of landing a blow on Ludovic. If it does, it inflicts 2d8 + 2 damage.


If he Dodges, then moves away, Ludovic subjects himself to an opportunity attack from the bugbear, but he also imposes disadvantage on the attack roll. Disadvantage reduces the bugbear’s chance to hit Ludovic from 50 percent to 25 percent, so Ludovic can expect to take 3 hp of damage. He’ll also make it only 30 feet away, so the bugbear will get to Attack him on its next turn, though it will have disadvantage on that roll, too—another 3 hp of damage, for a total of 6 hp. At the start of Ludovic’s next turn, he’s still within the bugbear’s reach.


If he Dashes, Ludovic subjects himself to an opportunity attack from the bugbear. The bugbear has a 50 percent chance to inflict 2d8 + 2 damage, so its expected damage is 6 hp—the same as if Ludovic had Dodged. However, if the bugbear uses its full movement to pursue Ludovic, it can’t catch up with him unless it uses its own action to Dash, too, in which case it will have no time to Attack. Therefore, Ludovic may have a decent chance of escaping.


If he Disengages, Ludovic gets to move away from the bugbear at his full movement speed without incurring an opportunity attack. However, if the bugbear uses its full movement to pursue Ludovic, it can catch up and then Attack him, doing expected damage of . . . 6 hp.

Well, crikey. It looks like no matter what Ludovic does, he’s going to end up taking 6 hp of damage from that bugbear! However, if he Dashes, he’s got the best chance of finishing his turn out of reach of the bugbear, so that’s one difference, at least.

But this scenario I’ve constructed is deceiving, because the bugbear just so happens to have an exactly 50 percent chance of hitting Ludovic in normal combat. Let’s take Ludovic out and substitute in two other characters: Dame Hella, a heavily armored paladin with AC 18; and Ffion, a characteristically speedy wood elf sorcerer with AC 11 and 35 feet of movement speed. The bugbear has only a 35 percent chance of damaging Hella in normal combat, for an expected 4 hp of damage; its chance of hitting Ffion, however, is 70 percent, for an expected 8 hp of damage.


Imposing disadvantage on the bugbear’s attack roll reduces the bugbear’s chance of hitting Hella from 35 percent to 12.25 percent, for expected damage of just 1 hp per hit or 3 hp for two (rounding, yo). The bugbear’s chance of hitting Ffion is reduced from 70 percent to 49 percent, for expected damage of 5 hp per hit or 11 hp for two.


The bugbear’s opportunity attack does an expected 4 hp of damage to Hella, 8 hp of damage to Ffion. Afterward, the bugbear can catch Hella only by Dashing, and it can’t catch Ffion at all—even if it Dashes too, she’s still got a 10-foot lead on it.


The bugbear can run down Hella and make its normal Attack, doing an expected 4 hp damage. But Ffion will have slipped out of its reach, so she’s at no risk.

Now we start to see the real differences between the three retreat options: Dodge is a good idea for combatants with high AC and a bad idea for combatants with low AC. Dash offers the best chance of escape, regardless of speed or armor class, at the cost of one round’s worth of combat damage. Disengage won’t reduce damage from a persistent opponent, but it does offer a faster combatant a chance to get away completely damage-free.

All this assumes one-on-one combat. If you have allies in the area, these actions open up more options. If your AC is high, Dodging can waste your opponent’s time while your allies bombard it with missile weapons and/or spells. Dashing past an ally means that any would-be pursuer may also be subject to one or more opportunity attacks. And Disengaging always makes sense if there’s a less injured or more durable ally willing to step in and serve as your rearguard. Conversely, if your opponent has allies, and you’re engaged in melee with more than one of them, Dashing away may subject you to multiple opportunity attacks, making Dodging or Disengaging more attractive.


So what can we conclude about what kind of creature or character Dodges, what kind Dashes, and what kind Disengages?

  • A creature that truly wants more than anything else to get away will Dash. It will probably take a hit on the way out, but only one—unless it’s engaged with multiple melee opponents.
  • An intelligent, disciplined creature that has backup, a high movement speed and/or a good measure of confidence that its opponent won’t pursue will Disengage.
  • A creature that’s especially hard to hit—because of armor, Dexterity or both—will Dodge, especially if what it really wants is to prolong the combat so that its allies can land additional hits, at minimal risk to itself.

Next: Drow.

This article has 6 comments

  1. JAMalcolmson Reply

    Your expected damage calculations imply hit probability affects damage somehow. This is only the case regarding the entirety of the battle, not the effects of one round. In the case of a single round, for your exmple, expected damage is always 4-18, with an average of 10.

    • Keith Ammann Reply

      No, that’s not right. Hit probability does affect damage, because you do no damage when you don’t hit. Expected damage per round equals average damage per hit times probability to hit, whether you’re looking at the whole battle or just one round at a time. You always have to account for the likelihood of a miss. Otherwise you’d expect an Archery specialist (who gets +2 to hit with ranged weapons) to do the same damage in any given round as an acrobat who happens to have the same Dexterity but who’s not even proficient with a bow. The Archery specialist has greater expected damage because it’s more likely that he’ll hit.

      The D&D 5E rules for calculating a monster’s challenge rating do assume that it hits every round, for average damage, but that’s not what we’re doing here.

  2. JAMalcolmson Reply

    I agree, you have to account for the probability of a miss. You already have by providing that probability. Folding that probability into expected damage doesn’t make any sense here though, because it doesn’t represent anything in actual play.

    A single hit from that thing could still one-shot many low level characters, regardless the “expected” damage calculation given here, or the improbability of that hit. Your expected damage calculation is only relevant over the course of multiple iterations for the purposes of determining how threatening the creature is likely to be in full combat, as it would represent the likely total damage output generated in proportion to misses across a given number of rounds. A single iteration isn’t a large enough range for it to mean anything. However, if you calculated out Hella’s hit probability and average damage against the bugbear and found the most likely number of rounds it would take for her to kill it, then multiplied the expected damage by the number of expected rounds, you will find out whether or not its expected damage output would be expected kill her in that time.

    In a single instance thoigh, if Hella takes the dodge action and the enemy hits, despite the low probability, the result will not be represented by the expected damage calculation at all- his chances of dealing 18 damage on a hit remain unaffected in this single instance. Theoretical work is only useful if it can model real things.

    • Keith Ammann Reply

      What I’m modeling, in this case, is the intuition gained as a result of experience, which is iterative. What I mean is, over the course of many encounters, we gain a sense of what’s more effective and what’s less effective, and probability is part of that. If I have two attacks, one of which can do massive nova damage but misses a lot of the time, while the other does modest but reliable damage, if I’m in a pinch, I can’t necessarily assume that I should go for the nova damage just because the average number is bigger. To-hit probability is relevant as well, because I may need the certainty of a hit more than I need to do a lot of damage at once, and by multiplying the average damage of each attack by its probability to hit, I get a sense of which direction I should err in when I’m not sure.

  3. Pingback: Goblinoids Revisted - The Monsters Know What They’re Doing

  4. Pingback: NPC Tactics: Swashbucklers and Master Thieves - The Monsters Know What They’re Doing

Leave a Comment

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