The conquest of Canaan at full strength — and the question of whether the God of the gentle portrait survives contact with it.
Written from inside the Christian confession, this study reads the Canaanite conquest at full strength and concedes plainly that the texts depict God commanding the Hebrew herem — devotion-to-destruction — with children named among the slain. It tests every honest mitigation in turn — archaeology, hyperbole, delayed judgment, the open door of mercy, a self-binding God who hands justice to human hands, the record kept as witness — marking exactly how far each reaches and where it fails. It lands more legible and no more acquitted: the framing illuminates God's posture but never justifies the commanded killing of infants, leaving the hardest verse a debt acknowledged, not paid.
This study is written from inside the Christian confession: we read these texts as Scripture, hold that the God they describe is good, and refuse to walk away from the canon when it turns hard. That is the claim now on trial. The companion portrait, What God Is Like, drew a God who is merciful, patient, and self-restraining. Here that portrait meets the texts that strain it past any other — the commanded destruction of the Canaanites, children included.
The strongest challenge is not subtle, and we will not blunt it: read plainly, these texts depict God commanding genocide. A faith that flinches from that sentence has not read its own book. So we will state it at full force, test every mitigation against it, mark each one for exactly how far it reaches and where it fails — and tell you honestly what is left when the testing is done. If a mitigation only goes partway, it will say partial on it.
This study also tests a specific reading — that the conquest belongs to a longer story beginning at the flood: a God who once judged the world directly, grieved it, bound himself against ever doing it that way again, thereafter let judgment run through human hands and covenant, held his own people to the same standard, and had the whole tragic record — his own grief included — written down as a witness. It is offered as interpretation, not the text's stated claim, and held to the same standard as everything else: where it illuminates, and where it fails.
Part 1 laid the road: the marked line of Canaan, the four-hundred-year clock, the wall around Israel, the long formation in the wilderness, and the archaeological doubt about whether the campaign happened as written. This part does the thing that road was leading to. It reads the commands.
There is no honest way to ease into this. The commands are explicit, repeated, and total.
"You must utterly destroy them… You shall make no covenant with them and show them no mercy."
Deuteronomy 7:2 · NRSVue
"You must not let anything that breathes remain alive. You shall annihilate them… so that they may not teach you to do all the abhorrent things that they do for their gods."
Deuteronomy 20:16–18 · NRSVue
The operative word is devote to destruction (חֵרֶם | ḥērem | ) — a term for total, irrevocable consecration by destruction. To place something under ḥerem is to hand it wholly over to God by wiping it out; it is the same root used when forbidden plunder must be burned rather than kept. Applied to a city, it means no survivors and no spoils. The narrative reports it carried out:
"Then they devoted to destruction by the edge of the sword all in the city, both men and women, young and old, oxen, sheep, and donkeys."
Joshua 6:21 · NRSVue (the fall of Jericho)
And then the hardest verse this study has to face — the command to Saul against Amalek, where the targets are named so that no reader can soften them:
"Now go and attack Amalek… do not spare them, but kill both man and woman, child and infant, ox and sheep, camel and donkey."
1 Samuel 15:3 · NRSVue
"Child and infant." The text does not let us pretend the violence stopped at combatants. Whatever else is said in this study, it is said after that verse and never around it.
State it plainly, because it deserves to be stated: this is divine-sanctioned genocide. A god who commands the killing of infants is, on the face of it, not good — and no appeal to that god's other qualities can be allowed to quietly cancel the plain sense of the order. The believer who rushes past these verses to the comforting ones is doing exactly what the site's own method forbids: letting theology overrule the text. The critic who stops here is reading honestly. Any response that does not begin by conceding the weight of this has already lost the argument.
That is the charge. The rest of this study is the attempt to answer it without cheating — and the willingness to report where the answer runs out.
The first thing an honest reader has to hold is that the conquest may not have occurred as Joshua narrates it. As Part 1 set out, the mainstream archaeological picture does not show a swift external blitzkrieg: Jericho's major destruction dates centuries too early on Kathleen Kenyon's excavation, Ai (et-Tell) shows an occupation gap at the relevant time, and the Merneptah Stele already places "Israel" inside Canaan. Many scholars read early Israel as having emerged largely from within Canaanite society rather than having destroyed it from outside.1
There is also strong evidence that the totalizing language is, in part, a genre. The Mesha Stele — a ninth-century Moabite inscription — has King Mesha "devoting" an Israelite town to his god Chemosh using the very same root, ḥrm, in the same breath as boasts of total slaughter. Across the ancient Near East, "I destroyed them utterly, I left none alive" was conventional victory rhetoric, often written by kings whose enemies demonstrably survived. Read in that world, "leave nothing that breathes alive" may be hyperbole for decisive defeat, the way a modern coach says a team was "annihilated."2
If the campaign is largely later narrative, then no historical genocide sits behind these chapters — a real relief, and a serious scholarly position. And if "utterly destroy" is conventional exaggeration, the body count on the ground was likely far smaller than the words imply.
But this relocates the problem; it does not remove it. A text that commands and commends total destruction as the will of God is a theological difficulty whether or not the swords ever fell that way — and worse in some respects, because now the killing is held up as an ideal to remember rather than an atrocity to regret. The hyperbole reading also thins precisely where it is needed most: "kill both man and woman, child and infant" (1 Sam 15:3) is hard to hear as a sportscaster's flourish. The archaeology buys real ground. It does not reach the children.
Set the historical question aside and take the text at its hardest — as a real command, meant as written. What can be said in the God-is-good direction? A great deal, and none of it sufficient. Here is the case, each piece tagged for its reach.
The text is emphatic that this is not about Israelite superiority: "Not because of your righteousness… but because of the wickedness of these nations" (Deut 9:4–5). The stated ground is accumulated practice — including child sacrifice: "they would even burn their sons and daughters in the fire to their gods" (Deut 12:31). There is real, if contested, archaeological corroboration that infant sacrifice existed in the broader Canaanite-Phoenician-Punic world — the urns of the Carthaginian tophet — though whether it was systematic or how widespread remains debated.3 Reach: this reframes the conquest as judicial rather than racial, which matters. Limit: "they did terrible things" is an argument about adults and a culture; it is not an argument for killing their infants, who did none of it.
"The iniquity of the Amorites is not yet complete" (Gen 15:16) frames the conquest as patience exhausted, not patience absent — four hundred years of waiting first. Reach: this fits the patient God of the portrait and rules out a capricious, trigger-quick deity. Limit: a long fuse on a judgment that finally includes children is still a judgment that finally includes children. Delay speaks to God's temperament; it does not answer the moral objection to the act itself.
The conquest is not airtight ethnic extermination, and the text itself breaks its own totalizing frame. Rahab, a Canaanite of Jericho, is spared with her whole household and enters the line of David and of Jesus (Josh 6:25; Matt 1:5). The Gibeonites secure their lives by treaty (Josh 9). Whoever turned toward Israel's God could live. Reach: this is the strongest single thread for the portrait's "invites even in judgment" theme — the boundary was permeable to repentance and was not finally about blood. Limit: the door being open to those who turned does not address those who never had the standing or the chance to turn — again, the children.
The same corpus that says "destroy" also, repeatedly, says "drive out" — dispossession, not annihilation: "little by little" (Deut 7:22; Exod 23:28–31), and Judges then reports Canaanites simply remaining in the land (Judg 1; 3:1–6). The two registers sit side by side, which is part of why many read the ḥerem language as rhetorical rather than a literal demographic program. Reach: it undercuts the flat "genocide manual" reading. Limit: the harshest verses are still on the page, still framed as divine speech, unexplained by the gentler ones.
Michael S. Heiser argued the ḥerem commands cluster on the territories of giant clans — the Anakim "from the Nephilim" (Num 13:33), Og "the last of the Rephaim" — and so were aimed at a specific, mythologically-charged bloodline tied to a cosmic rebellion (Deut 32:8–9; Ps 82), not at ordinary Canaanites, who could be driven out or spared.
We engage it because it does real work on the "why these peoples, why total" question — and we decline to lean on it. It stitches together disparate texts, the mainstream does not accept the link, and it cannot reach the verse that matters most: Amalek descends from Esau, not the Nephilim, yet "child and infant" is commanded there too. A framework that leaves the hardest case untouched is not the answer.
An old and serious move holds that God, as the author and giver of every life, has a right over its end that no human commander has — so a divinely ordered death is judgment (the penalty God may justly impose), not murder (a moral evil). On this view God authors malum poenae, the penalty, never malum culpae, moral wrong. Reach: it is internally coherent within the theistic frame and is the backbone of most careful defenses. Limit: grant God the prerogative over life and death, and you have answered whether God may take a life; you have not answered why the taking is done by human soldiers with swords, by command, against infants, in a way that would be monstrous in any other mouth. The prerogative explains the dying. It does not explain the order to a man to do the killing. That gap is where the defense, honestly, does not close.
Step back to the first time the Bible shows God facing a world gone wrong on a global scale. He does not strike in cold prerogative. He grieves:
"And the LORD was sorry that he had made humankind on the earth, and it grieved him to his heart."
Genesis 6:6 · NRSVue
The verb is was sorry / grieved (נָחַם | nāḥam | ) — the same root the Bible uses for relenting and for being moved to compassion. The flood is judgment soaked in regret, not relish. And on the far side of it, God does something he never undoes: he binds himself.
"I will never again curse the ground because of humankind, for the inclination of the human heart is evil from youth; nor will I ever again destroy every living creature as I have done."
Genesis 8:21 · NRSVue
Read the reason carefully, because it is startling: not "because humanity has improved" — the very same evil inclination is named — but because God will hold himself back. The bow set in the cloud (Gen 9:13–17) is the sign of a self-imposed limit. From here on, the world is run by a God who has sworn off the total reset. And in the same breath, the administration of justice is handed to human hands:
"Whoever sheds the blood of a human, by a human shall that person's blood be shed."
Genesis 9:6 · NRSVue
This is the hinge the reading turns on, and it is a real feature of the text, not an imposition: after the flood, judgment shifts from God-as-universal-destroyer toward judgment carried through covenant, law, and human agents. The conquest, on this reading, is not a capricious deity's bloodlust; it is judgment delegated by a God who has bound himself against doing it himself, wholesale, ever again.
The "handover" is not clean, and this study will not pretend it is. God keeps acting directly after the flood — fire on Sodom (Gen 19), the plagues and the sea (Exodus), the ground swallowing Korah (Num 16). So the self-binding is best read narrowly: against the total, world-erasing reset, not against all direct judgment. The shift toward human agency is real but partial, and stated as a clean rule it overreaches the text.
If the conquest were ethnic favoritism — our god clearing land for us — you would expect Israel to be exempt from the standard it enforces. They are not. The very chapter that catalogs the Canaanites' abominations turns and aims the identical threat at Israel:
"…otherwise the land will vomit you out for defiling it, as it vomited out the nation that was before you."
Leviticus 18:28 · NRSVue
The ḥerem-logic is not a weapon Israel owns; it is a standard Israel is placed under — and the Old Testament makes good on it, driving Israel into exile when it does what the Canaanites did. Whatever else this is, it is not a tribal deity playing favorites. And the posture behind the judgment, read across the canon, is reluctance, not eagerness:
"As I live, says the Lord GOD, I have no pleasure in the death of the wicked, but that the wicked turn from their ways and live. Turn back, turn back from your evil ways; for why will you die, O house of Israel?"
Ezekiel 33:11 · NRSVue
"How can I give you up, Ephraim?… My heart recoils within me; my compassion grows warm and tender" (Hosea 11:8). Judgment is called God's "strange" and "alien" work (Isaiah 28:21) — foreign to his nature. "He does not willingly afflict or grieve anyone" (Lamentations 3:33). And at the far end, the Son weeps over the city before he names its ruin (Luke 19:41). This God pleads — repeatedly, at length — with the very people he warns, begging them not to become what judgment falls on.
Here is the honest catch, and it is load-bearing. Nearly all of this recorded grief is over Israel, and nearly all of it is in the prophets — not in the conquest narratives themselves. The book of Joshua does not show God weeping over Jericho. The grief is a real, pervasive biblical theme, but applying it to the Canaanites' destruction is an inference drawn from how God grieves elsewhere; it is not depicted on the page where the Canaanites die. A reading that leans on God's grief at the conquest is borrowing it from texts about other judgments. That borrowing may be fair — but it is borrowed, not given.
The flood hinge (§IV) folded judgment-in-human-hands into the defense: a self-bound God runs justice through covenant and human agents. The same book that hands over the sword also shows, without flinching, what a human hand does with it. Return to the verse this study named as the residue — the command against Amalek, "kill both man and woman, child and infant, ox and sheep, camel and donkey" (1 Sam 15:3). Saul is given that order and refuses it: he spares Agag and the best of the livestock (15:9). Seven chapters later, enraged that the priests of Nob had unwittingly helped David, the same Saul turns the same formula on a town of his own people:
"Nob, the city of the priests, he put to the sword; men and women, children and infants, oxen, donkeys, and sheep, he put to the sword."
1 Samuel 22:19 · NRSVue
The echo lands where it is cruelest: the merism child and infant (עוֹלֵל וְיוֹנֵק | ʿôlēl wᵉyônēq | ) stands in both 15:3 and 22:19, and the livestock list mirrors Jericho's (Josh 6:21). The man who would not devote God's enemy to destruction devotes God's own priests to it — and nobody commanded him. His own guards refuse the order; only Doeg the Edomite, a foreigner, will carry it out (22:17–18).
Nob is not technically a ḥerem. The verse uses "struck… with the edge of the sword" (nakah + lᵉpî-ḥereb), not the consecration-root ḥrm; there is no altar-logic and no divine word behind it. What ties it to the conquest is the annihilation-formula — the same total victim-list, infants included — not the technical vocabulary. The parallel is the narrator's, drawn by the shared words; it is shown, not a claim the text spells out. (The same massacre is read for its effect on the priestly oracle in Forfeited Access; here only its moral weight is in view.)
It confirms one defense and wounds another. It confirms non-exemption (§V) at narrative full strength: the baby-killing formula falls on Israelites, by an Israelite king, and the book records it among the darkest acts of his reign. The text plainly does not reserve the horror for outsiders.
But it exposes the soft joint in the flood hinge. If the post-flood order runs judgment "through human hands," Nob is what that delegation looks like when the hand is a furious king: uncommanded slaughter of the innocent, dressed in the very words Scripture elsewhere puts in God's own mouth. One honest complication: the killing intersects the foretold fall of Eli's priestly house (1 Sam 2:31–33), so a reader can ask whether it is divine judgment after all. But the text never has God order it; Saul acts from rage, and Scripture treats him — like Assyria in Isaiah 10:5–15 — as the guilty instrument of a judgment he was never authorized to execute. Foreknowledge is not a warrant.
That the same Bible can command the formula against Amalek and damn a man for using it on Nob — with the same words — is exactly the residue this study refuses to dissolve. It shows the author is not morally numb to the formula, and that the formula is monstrous in a human mouth. It still does not tell us why the command in 15:3 is not.
There is an objection that has to be met head-on: if the purpose of the killing was to teach Israel — to make them feel the weight of judgment — then the Canaanite children become instruments of someone else's moral education, which is worse, not better. That objection lands against one version of the reading. It does not land against this one, and the difference is the whole point.
The killing's stated purpose is judgment. The recording of it — the grief, the warnings, the reluctance, the cost — has a different purpose, and the text claims it outright:
"These things happened to them to serve as an example, and they were written down to instruct us, on whom the ends of the ages have come."
1 Corinthians 10:11 · NRSVue
"Whatever was written in former days was written for our instruction" (Romans 15:4); the Song of Moses is given expressly as "a witness… it will not be lost from the mouths of their descendants" (Deuteronomy 31:19–21). On this reading the children are not killed so that we learn; the deaths are judgment, and the tragedy is recorded — God's grief included — so that everyone afterward can see it. The victims are not means to a lesson. The text is the witness; we are the ones meant to read it and not repeat it.
Reach: this genuinely answers the instrumentalization objection. The pedagogy lives in the recording, not in the deaths, and Scripture explicitly claims that function for itself. It also fits the flood hinge — a God who grieves and self-limits is exactly the kind to preserve a painful record rather than sanitize it.
Limit: a witness explains why the story is told; it does not make the act just. That the children's deaths are recorded as testimony does not establish that commanding them was right — and the reach depends on the grief actually being in the record, which, as the prior seam shows, is thinner at the conquest than the reading needs.
Lay it all out — the standard defenses with their honest limits, the flood hinge, the non-exemption, the pleading God, the record kept as witness. Something substantial is achieved: the conquest stops looking like a tribal deity's bloodlust and starts looking like one dark stretch of a long, coherent story — a God who judges reluctantly, binds himself, hands justice to human hands, holds his own people to the same standard, and preserves the whole grieving record so it will be seen. That is a far more answerable God than the cartoon the indictment attacks. It is, in fact, a lot.
And it is still not enough — because one thing stands exactly where it stood before any of it was said:
"Do not spare them, but kill both man and woman, child and infant."
1 Samuel 15:3 — the residue no framing reachesEvery move in this study explains something around the commanded death of children — why judgment runs through human hands, why it is not favoritism, why it is grieved, why it is written down. None of them explains the command itself. The framing illuminates God's posture; it does not establish the act's justice. A grieving, self-limiting, record-keeping God who nonetheless orders the killing of infants is more coherent as a character and no less troubling as a fact. We hold the verse; we do not hold a clean answer to it.
Tested through the flood-and-witness reading, the good-God claim comes out more legible and no more acquitted. We can now say a great deal about why the conquest looks the way it does and how it sits in the larger story — a self-bound God, judgment in human hands, the same standard turned on Israel, the grief, the record kept on purpose. What we still cannot say is that commanding the death of the children was right. The gain is real; the residue is exactly the size it always was.
One honesty owed to reader and author both: this is the framing the author finds most persuasive, which is exactly why its weakest joint — that the conquest's grief is borrowed from the prophets, not shown in Joshua — has been marked rather than smoothed. What keeps the whole from being unfalsifiable is the one stake the faith leaves exposed: if Christ is not raised, the structure fails (1 Corinthians 15:14). Short of that, the believer trusts the shape of the story while holding the hardest verse as a debt acknowledged, not paid.
That is where it lands — not an acquittal that pretends there was no charge, and not a conviction that throws out the book. The God of the portrait is still standing at the end of the worst text, and we can describe his bearing there more fully than before: grieved, self-limited, holding his own people to the same measure, keeping the record so it cannot be forgotten. He is still standing over ground he has not explained. The honest believer stands there with him, uneasy, and stays.
Sources
The Holy Bible, New Revised Standard Version, Updated Edition (NRSVue), 2021.
Position: Mainline Protestant translation committee; ecumenical, critically based text.
Strong's / Blue Letter Bible — Hebrew ḥērem / ḥāram (H2763), "devote to destruction."
Position: Reference lexicon; non-confessional lexical data.
Copan, P., & Flannagan, M. (2014). Did God Really Command Genocide? Baker Books.
Position: Conservative evangelical — defends the hyperbole/dispossession reading and the judgment framing.
Younger, K. L., Jr. (1990). Ancient Conquest Accounts. JSOT Press.
Position: Conservative evangelical / ANE comparative — conquest totalizing language as a shared genre convention.
Seibert, E. (2009). Disturbing Divine Behavior. Fortress Press.
Position: Mainline Protestant — argues the conquest texts cannot be harmonized with the God revealed in Christ; "the textual God" vs. "the actual God."
Boyd, G. A. (2017). The Crucifixion of the Warrior God. Fortress Press.
Position: Evangelical (non-traditional) — divine self-limitation and a cruciform/Christ-centered rereading of OT violence; engaged for the self-binding trajectory, not adopted wholesale.
Finkelstein, I., & Silberman, N. A. (2001). The Bible Unearthed. Free Press.
Position: Historical-critical / archaeological — conquest largely non-historical as narrated.
Heiser, M. S. (2015). The Unseen Realm. Lexham Press.
Position: Conservative evangelical (Semitic languages) — divine-council / Nephilim reading of the ḥerem; engaged here, not adopted.
Mesha Stele (Moabite Stone), c. 840 BCE, Louvre.
Position: Primary artifact — cognate ḥrm usage; ANE conquest rhetoric.