Checked against Palworld v1.0.3 · verified 2026-08-25 · Data status
Breeding mechanics
Palworld 1.0 decides what hatches from a pair by breeding rank, not by element or by type: the target is floor((rankA + rankB + 1) / 2), and the child is the eligible Pal whose own rank sits closest to it. Replayed against this dataset the rule reproduces 99.2% of ordinary pairs, with 79 fixed combinations it is not meant to predict at all.
Palworld decides what hatches from a pair by rank, not by element or by type. Every Pal has a breeding rank, and the pair's target rank is floor((rankA + rankB + 1) / 2) — the two parents' ranks averaged, rounded up. The child is the Pal whose own rank sits closest to that target.
Closest is doing real work in that sentence: the target rarely lands exactly on a Pal, so the rule picks the nearest one, and a tie goes to the higher breeding rank. The search is also not over every Pal. 194 of them are eligible to be picked this way; the rest are excluded because they come only from a fixed combination, because they breed only from themselves, or because nothing breeds them at all.
Replayed against this dataset, the rule reproduces 44117 of the 44473 ordinary pairs in Palworld 1.0, 99.20% of them. The remaining 356 are pairs the game resolves some other way, and 79 combinations are explicit overrides the rule is not meant to predict at all.
Worked through: Panthalus has rank 20 and Shaolong has rank 40, so the target is (20 + 40 + 1) / 2 rounded down, which is 30. The eligible Pal nearest that number is Aegidron at rank 30 — and that is the pair's child in the dataset, verified against the same edge every other page on this site reads.
This is worth knowing and worth not relying on. Working the arithmetic out by hand tells you why a pair produces what it produces, but the combination pages on this site are lookups against the recorded pairs, not applications of this rule — which is why they are right about the overrides too. If the two ever disagree, the recorded pair is the answer.
The rank rule decides the species only. It has nothing to say about which passive skills, active skills or stats the offspring gets: those are separate systems with separate rules, and whether a specific build on the resulting Pal can exist is the question the build checker answers.
Data from Palworld v1.0.3, verified 2026-08-25.
Those 356 pairs are resolved by something other than rank arithmetic in Palworld 1.0 — a fixed rule, or a constraint the published rank values do not capture. The combination pages are lookups against the recorded pairs rather than applications of this rule, which is why they are right about these too.
No. The rank rule decides the species and nothing else. Passive skills, active skills and stats are separate systems with separate rules, and whether a specific build on the resulting Pal can exist at all is the question the build checker answers.
Under the rank rule, no: two Pals of one species average to their own rank, so the nearest eligible Pal is that species again. The exceptions are the 79 explicit overrides, which are listed on the combination page of the Pal they produce.
Working the arithmetic by hand tells you why a pair produces what it produces; the calculator answers the same question against the recorded pairs, including the overrides the rule cannot predict.