diff --git a/wikify/engine/store.py b/wikify/engine/store.py
index 72d154c..1957162 100644
--- a/wikify/engine/store.py
+++ b/wikify/engine/store.py
@@ -11,6 +11,10 @@
 import frappe
 from frappe.utils.file_manager import save_file
 
+# `Data` column limit — titles come out of PDFs and LLMs, so their length is unbounded
+# and `insert()` throws rather than truncating.
+TITLE_MAX = 140
+
 
 def create_document(
 	title: str,
@@ -21,7 +25,7 @@ def create_document(
 ) -> str:
 	"""Create a Source Document and return its name."""
 	doc = frappe.new_doc("Source Document")
-	doc.title = title
+	doc.title = title[:TITLE_MAX]
 	doc.set("import", import_name)  # 'import' is a Python keyword — set by string
 	doc.project = project  # denormalized from the Import for project-scoped Explore
 	doc.pdf = pdf_url
@@ -331,7 +335,7 @@ def replace_sections(source_document: str, sections) -> int:
 		doc.source_document = source_document
 		doc.parent_source_section = path_to_name.get(tuple(sec.hierarchy_path[:-1]))
 		doc.is_group = 1 if tuple(sec.hierarchy_path) in parent_paths else 0
-		doc.title = sec.title
+		doc.title = sec.title[:TITLE_MAX]
 		doc.section_type = sec.section_type
 		doc.level = sec.level
 		doc.hierarchy_path = " > ".join(sec.hierarchy_path)
diff --git a/wikify/tests/test_sectionize.py b/wikify/tests/test_sectionize.py
index 7c48535..84828b0 100644
--- a/wikify/tests/test_sectionize.py
+++ b/wikify/tests/test_sectionize.py
@@ -8,9 +8,9 @@
 import frappe
 from frappe.tests.utils import FrappeTestCase
 
-from wikify.engine import parse_pdf, remediate_pdf
+from wikify.engine import parse_pdf, remediate_pdf, store
 from wikify.engine.loader.cleanup import clean_pages, strip_outer_markdown_fence
-from wikify.engine.loader.sectionizer import sectionize
+from wikify.engine.loader.sectionizer import Section, sectionize
 from wikify.tests.test_parse_pipeline import _make_sample_pdf
 from wikify.tests.test_remediate_pipeline import _MERMAID, _fake_chat
 
@@ -136,6 +136,24 @@ def test_clean_pages_keeps_data_row_mentioning_approved_by_once(self):
 		self.assertIn("|---|---|", cleaned[1])  # the real table separator survives
 
 
+class TestTitleFit(FrappeTestCase):
+	def test_overlong_section_title_is_stored_truncated(self):
+		long_title = "A. " + "Responsibilities of the transplant coordinator " * 8
+		self.assertGreater(len(long_title), store.TITLE_MAX)
+		sd = store.create_document("Title Fit Test")
+		section = Section(
+			title=long_title, level=1, hierarchy_path=[long_title], page_start=1, page_end=1, markdown="body"
+		)
+		store.replace_sections(sd, [section])
+
+		stored = frappe.get_all(
+			"Source Section", filters={"source_document": sd}, fields=["title", "markdown"]
+		)
+		self.assertEqual(len(stored), 1)
+		self.assertEqual(stored[0].title, long_title[: store.TITLE_MAX])
+		self.assertEqual(stored[0].markdown, "body")
+
+
 class TestSectionizeIntegration(FrappeTestCase):
 	"""parse/remediate build the Source Section NestedSet tree."""