diff --git a/chrome/content/zotero/xpcom/duplicates.js b/chrome/content/zotero/xpcom/duplicates.js
index 56f4812a8e..dafa57f760 100644
--- a/chrome/content/zotero/xpcom/duplicates.js
+++ b/chrome/content/zotero/xpcom/duplicates.js
@@ -211,6 +211,8 @@ Zotero.Duplicates.prototype._findDuplicates = async function () {
 			let row = rows[i];
 			let newVal = Zotero.Utilities.cleanISBN('' + row.value);
 			if (!newVal) continue;
+			// Canonicalize to ISBN-13 so an ISBN-10 and its ISBN-13 equivalent match
+			newVal = Zotero.Utilities.toISBN13(newVal);
 			isbnCache[row.itemID] = newVal;
 			newRows.push({
 				itemID: row.itemID,
@@ -221,6 +223,28 @@ Zotero.Duplicates.prototype._findDuplicates = async function () {
 		processRows(newRows);
 	}
 	
+	// Also cache ISBNs of book sections. Their ISBN identifies the containing book rather
+	// than the section itself, so book sections aren't unioned directly by ISBN, but the
+	// cached value is still used below to veto false title/creator matches.
+	var sql = "SELECT itemID, value FROM items JOIN itemData USING (itemID) "
+				+ "JOIN itemDataValues USING (valueID) "
+				+ "WHERE libraryID=? AND itemTypeID=? AND fieldID=? "
+				+ "AND itemID NOT IN (SELECT itemID FROM deletedItems)";
+	var rows = await Zotero.DB.queryAsync(
+		sql,
+		[
+			this._libraryID,
+			Zotero.ItemTypes.getID('bookSection'),
+			Zotero.ItemFields.getID('ISBN')
+		]
+	);
+	for (let i = 0; i < rows.length; i++) {
+		let row = rows[i];
+		let newVal = Zotero.Utilities.cleanISBN('' + row.value);
+		if (!newVal) continue;
+		isbnCache[row.itemID] = Zotero.Utilities.toISBN13(newVal);
+	}
+	
 	// DOI
 	var sql = "SELECT itemID, value FROM items JOIN itemData USING (itemID) "
 				+ "JOIN itemDataValues USING (valueID) "
diff --git a/test/tests/duplicatesTest.js b/test/tests/duplicatesTest.js
index d86d382de5..e10aeee483 100644
--- a/test/tests/duplicatesTest.js
+++ b/test/tests/duplicatesTest.js
@@ -110,4 +110,60 @@ describe("Duplicate Items", function () {
 			assert.sameMembers(item3.relatedItems, [item1.key]);
 		});
 	});
+
+	describe("ISBN matching", function () {
+		async function getDuplicateSets() {
+			var duplicates = new Zotero.Duplicates(Zotero.Libraries.userLibraryID);
+			var search = await duplicates.getSearchObject();
+			return search.search();
+		}
+
+		it("should match books with equivalent ISBN-10 and ISBN-13", async function () {
+			var item1 = await createDataObject('item', { itemType: 'book', title: 'Effective Java' });
+			item1.setField('ISBN', '0134685997');
+			await item1.saveTx();
+
+			var item2 = await createDataObject('item', { itemType: 'book', title: 'Effective Java, 3rd Edition' });
+			item2.setField('ISBN', '9780134685991');
+			await item2.saveTx();
+
+			var ids = await getDuplicateSets();
+			assert.include(ids, item1.id);
+			assert.include(ids, item2.id);
+		});
+
+		it("should not match books with different ISBNs and titles", async function () {
+			var item1 = await createDataObject('item', { itemType: 'book', title: 'Clean Code' });
+			item1.setField('ISBN', '0062316095');
+			await item1.saveTx();
+
+			var item2 = await createDataObject('item', { itemType: 'book', title: 'Learning JavaScript' });
+			item2.setField('ISBN', '0596009089');
+			await item2.saveTx();
+
+			var ids = await getDuplicateSets();
+			assert.notInclude(ids, item1.id);
+			assert.notInclude(ids, item2.id);
+		});
+
+		it("should not match book sections with the same title/creator but different container ISBNs", async function () {
+			var item1 = await createDataObject(
+				'item',
+				{ itemType: 'bookSection', title: 'Introduction', creators: [{ lastName: 'Smith', firstName: 'John', creatorType: 'author' }] }
+			);
+			item1.setField('ISBN', '0123456789');
+			await item1.saveTx();
+
+			var item2 = await createDataObject(
+				'item',
+				{ itemType: 'bookSection', title: 'Introduction', creators: [{ lastName: 'Smith', firstName: 'John', creatorType: 'author' }] }
+			);
+			item2.setField('ISBN', '0987654322');
+			await item2.saveTx();
+
+			var ids = await getDuplicateSets();
+			assert.notInclude(ids, item1.id);
+			assert.notInclude(ids, item2.id);
+		});
+	});
 });