I know I promised yugoloths today, but there are a couple of things I omitted in my discussion of demon tactics. One was the quasit, a weak demon that’s easily summoned and that occasionally even serves as a wizard’s familiar (an evil wizard, one would expect—either that or one with poor judgment). The other is the variant “Demon Summoning” rule (Monster Manual, page 54).

The quasit is a tiny fiend with very low Strength, very high Dexterity and average Constitution. This ability profile would normally suggest a sniper, but the quasit has no ranged attack. Therefore, this is a creature that has to be able to repeatedly strike, then slip away. Maybe it doesn’t even initiate attack itself—maybe it waits in hiding, using its Invisibility feature, until its master or an ally is already engaged in combat with an enemy, then pops out to deliver sneak attacks from behind (“Optional Rule: Flanking,” Dungeon Master’s Guide, page 251—live it, love it!), or maybe it takes advantage of its high speed and Invisibility to ambush ranged attackers and spellcasters. Either of these would be consistent with its proficiency in the Stealth skill.

Quasits are none too bright: their Intelligence score of 7 indicates that they operate primarily by instinct. However, they can assess, in a very broad sense, whether an opponent is “tough” or “not tough.” Big targets, armored targets, and targets with large or heavy weapons are “tough.” Small targets, unarmored targets, and targets using small or simple weapons are “not tough.” Against a “tough” opponent, they lead with Scare, attempting to startle the opponent and throw it off before attacking with their claws or bite. Against a “not tough” opponent, they skip Scare and go directly to melee attacks.

If an opponent does any damage at all against a quasit, its next action is to turn invisible. It doesn’t attack again unless and until a potential opponent is otherwise engaged.

Quasits have the ability to assume the form of a bat, centipede or toad; each of these forms enhances its movement in some way, none of which is quite enough to make the tactic of maintaining distance in order to dart in, attack and dart back out workable. The risk of a successful opportunity attack is simply too great for a monster with armor class 13 and only 7 hp, even one with resistance to physical damage from normal weapons, and even if the enemy has disadvantage from being frightened or poisoned.


On to demon summoning. Once per day, various demons may attempt to summon backup; none of them has a 100 percent chance of success. Therefore, the use of an action to summon one or more other demons always has to be weighed against the chance of failure, which costs an entire turn’s ability to act. If the would-be summoner has a powerful Multiattack, the opportunity cost is multiplied.

  • A barlgura has a 30 percent chance of summoning one other barlgura. That’s spending one action in the hope of gaining, say, two or three extra, equivalent to those of the summoner. Not worth it, especially since there are probably multiple barlguras on the scene already.
  • A vrock has a 30 percent chance of summoning one other vrock or 2d4 dretches. The one-vrock option isn’t worth it, for the same reason it’s not worth it for one barlgura to summon another. As for the dretches, their Fetid Cloud is somewhat redundant to the vrock’s Spores, and with their low challenge rating of 1/4, even five of them don’t add up to a whole lot next to the vrock’s CR 6, even if their numbers do increase the difficulty of the encounter; my suggestion to other dungeon masters is that if you want dretches to be part of your vrock encounter, just put them there in the first place.
  • A chasme has a 30 percent chance of summoning one other chasme. Not worth it.
  • A hezrou has a 30 percent chance of summoning one other hezrou or 2d6 dretches. See vrocks, above.
  • A glabrezu has a 30 percent chance of summoning one other glabrezu, 1d2 hezrous or 1d3 vrocks. Forget the other glabrezu; not only is it not tactically worth it, they just aren’t the collaborative type. What about the hezrous or vrocks? Well, now we do begin to see that the addition of even a couple of extra enemies can have a real effect, thanks to their moderate challenge ratings. Hezrous could be useful for distracting or shutting down ranged attackers that are giving the glabrezu problems, but I think vrocks would be even better for this, because of their Spores, their Stunning Screech and their ability to fly. The low probability of succeeding in the summoning, however, means that this is not an ability that the glabrezu should wait until it’s in real trouble to use. I think a glabrezu would try to summon vrocks as soon as it realized it was in real trouble, which is to say, if its opponents managed to at least moderately injure it (reduce it to 109 hp or fewer) in the first turn of combat.
    Even so, let’s drill down to actual probabilities, because a glabrezu is capable of calculating these things. It has a 30 percent chance of summoning an average of two vrocks; let’s say combat will continue for three more rounds if it succeeds. The vrock’s powers grant them and the glabrezu advantage on all attacks. Spores has a 50 percent chance of dealing 5.5 hp of damage each to three opponents. On the two subsequent turns, each vrock has a 60 percent chance (against AC 15) of dealing 24 hp of damage with its Multiattack. Putting it all together, the summoning attempt can be expected to do an additional 14 hp of damage, plus the effects of advantage and disadvantage on whatever damage the glabrezu itself does. In contrast, the glabrezu, with a 75 percent chance to hit (93.75 percent with advantage) does an expected 37 hp of damage per turn against the same AC, all by itself. In other words, the total damage inflicted by the vrocks over multiple rounds—even accounting for their effect on the glabrezu’s own damage—doesn’t equal what the glabrezu can do itself in one turn. Sure, it will also reduce the damage the player characters are able to do to the glabrezu, but when you get right down to it, the likelihood of failure is so much greater than the likelihood of success that the glabrezu is going to err on the side of DIY. Final verdict: no summoning.
  • A yochlol is most likely to be encountered when summoned by a drow priestess of Lolth, and in that situation, it can’t summon other demons itself, so we can ignore this one.
  • With the nalfeshnee, the chance of successful summoning increases from 30 percent to 50 percent; we’re also looking at demons that are likely to have many minions behind them already, so they won’t be doing all the work themselves. They can afford to take a turn to bring more allies into the fight. Therefore, rather than calculating to-hit probabilities and expected damage, I’m going to base this assessment on the effect on the total experience points of the encounter. Summoning 1d4 vrocks will bring an average of 5,750 XP of new demon blood into the encounter; 1d3 hezrous, 7,800; 1d2 glabrezus, 7,500; and one more nalfeshnee, 10,000. None of these summonings is likely to increase the encounter multiplier. But consider something else: summoning one more nalfeshnee just brings a rival commander onto the field, so we can probably write that one off. Subtracting that option, the hezrous look like the preferred second choice.
  • Similarly, a marilith probably won’t want to summon another marilith—that’s just asking for a power struggle. Summoning 1d6 vrocks will add an average of 8,050 XP, with a 50 percent chance of raising the encounter multiplier, so keep that in the back of your mind. Summoning 1d4 hezrous will add an expected 9,750 XP; 1d3 glabrezus, 10,000 XP; 1d2 nalfeshnees, 15,000 XP. Raising the encounter multiplier one level increases the AXP by half the total XP value of all monsters in the encounter; for the marilith alone, that’s another 3,750 AXP. If the total XP of the marilith’s minions are equal to or greater than those of the marilith, the vrocks take pole position. Otherwise, stick with the nalfeshnees.
  • Finally, the balor. Summoning 1d8 vrocks adds an expected 10,350 XP and has a 67.5 percent chance of raising the encounter multiplier; 1d6 hezrous, 13,650 XP and a 50 percent chance of raising the multiplier; 1d4 glabrezus, 12,500 XP; 1d3 nalfeshnees, 20,000 XP; 1d2 mariliths, 22,500 XP; and one goristro, 18,000 XP. We can toss out the goristro right away—unless the balor’s enemies are hiding inside or behind a structure that a goristro could demolish. As for vrocks and hezrous, unless the balor has a formidable army behind it, it’s going to be hard for them to add enough AXP to the encounter to outweigh the effect of bringing two mariliths onto the field. I think the mariliths have it.

At this point, I want to throw in a caution to other DMs. One thing I as a DM really appreciate about fifth-edition Dungeons and Dragons is the tools it’s given us for calculating encounter balance. The more uncertainty we introduce into an encounter, the more likely our players will experience either an unsatisfying cakewalk or a demoralizing team wipe.

In every single one of the scenarios above, either a failed summoning or an unusually successful one (in which the boss demon summons the greatest possible number of minions) has the potential to tip an encounter way off balance. To be blunt, I think you shouldn’t use this rule as written. Instead, decide beforehand whether you want minions to appear at a certain point (or from the outset), then declare it to be so. Random summoning works in some instances, but I’ve concluded that this isn’t one of them.

Next: yugoloths, for realz this time.

Leave a Comment

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