diff --git a/backend/app/routers/products.py b/backend/app/routers/products.py
index bbcabfb..ae68d24 100644
--- a/backend/app/routers/products.py
+++ b/backend/app/routers/products.py
@@ -1,3 +1,6 @@
+import difflib
+import re
+
 from fastapi import APIRouter, BackgroundTasks, Depends, File, HTTPException, UploadFile, status
 from fastapi.responses import Response
 from sqlalchemy.orm import Session, joinedload, selectinload
@@ -9,6 +12,7 @@ from ..models import (
     Barcode,
     BaseUnit,
     Category,
+    CategoryTracking,
     Group,
     Location,
     Movement,
@@ -25,6 +29,9 @@ from ..schemas import (
     BarcodeCreate,
     LocationMinStockIn,
     LookupResult,
+    MatchCandidate,
+    MatchLine,
+    MatchRequest,
     ProductCreate,
     ProductOut,
     ProductUpdate,
@@ -35,6 +42,7 @@ from ..schemas import (
 from ..services import images
 from ..services.categories import suggest_category
 from .categories import descendant_ids
+from .settings import get_receipt_match_threshold
 from ..services.conversion import ConversionError, resolve_product_unit
 from ..services.fields import FieldError, apply_field_values
 from ..services.group_codes import detach as detach_group_code, sync as sync_group_code
@@ -42,6 +50,76 @@ from ..services.stock import removal_stats
 
 router = APIRouter(prefix="/products", tags=["products"])
 
+# ---------------------------------------------------------------------------
+# Kassenzettel-Abgleich: eine OCR-Zeile den ähnlichsten Lebensmitteln zuordnen.
+# ---------------------------------------------------------------------------
+_UMLAUTE = str.maketrans({"ä": "ae", "ö": "oe", "ü": "ue", "ß": "ss"})
+
+
+def _normalisieren(text: str) -> str:
+    """Klein, Umlaute aufgelöst, nur Buchstaben/Ziffern – Preise/Sonderzeichen weg."""
+    text = (text or "").lower().translate(_UMLAUTE)
+    return re.sub(r"[^a-z0-9]+", " ", text).strip()
+
+
+def _match_score(zeile: str, name: str) -> int:
+    """Ähnlichkeit 0–100 zwischen Kassenzeile und Produktname.
+
+    Kombiniert die Gesamt-Ähnlichkeit mit einem Token-/Präfix-Abgleich, damit
+    abgekürzte Kassennamen ("MÜHLEN SCHNITZ") auf den vollen Namen passen.
+    """
+    a, b = _normalisieren(zeile), _normalisieren(name)
+    if not a or not b:
+        return 0
+    gesamt = difflib.SequenceMatcher(None, a, b).ratio()
+    a_tok, b_tok = a.split(), b.split()
+    bester = []
+    for t in a_tok:
+        m = 0.0
+        for u in b_tok:
+            if u.startswith(t) or t.startswith(u):
+                m = max(m, min(len(t), len(u)) / max(len(t), len(u)))
+            else:
+                m = max(m, difflib.SequenceMatcher(None, t, u).ratio())
+        bester.append(m)
+    token = sum(bester) / len(bester) if bester else 0.0
+    return round(100 * max(gesamt, token))
+
+
+@router.post("/match", response_model=list[MatchLine])
+def match_receipt(
+    payload: MatchRequest,
+    db: Session = Depends(get_db),
+    _: User = Depends(get_current_user),
+) -> list[MatchLine]:
+    """Kassenzettel-Zeilen den ähnlichsten LEBENSMITTELN zuordnen (Score 0–100).
+    Gegenstände (auch Verbrauchsgegenstände) bleiben außen vor; je Zeile die besten
+    Treffer über dem Schwellwert (aus Request oder Einstellung)."""
+    schwelle = payload.threshold if payload.threshold is not None else get_receipt_match_threshold(db)
+    schwelle = max(0, min(100, schwelle))
+    lebensmittel = [
+        p
+        for p in db.query(Product).options(joinedload(Product.category)).all()
+        if product_tracking(db, p) != CategoryTracking.object.value
+    ]
+    ergebnis: list[MatchLine] = []
+    for zeile in payload.lines:
+        text = (zeile or "").strip()
+        treffer = [(p, _match_score(text, p.name)) for p in lebensmittel]
+        treffer = sorted(
+            (ps for ps in treffer if ps[1] >= schwelle),
+            key=lambda ps: ps[1],
+            reverse=True,
+        )
+        ergebnis.append(MatchLine(
+            text=text,
+            candidates=[
+                MatchCandidate(product_id=p.id, name=p.name, brand=p.brand, score=s)
+                for p, s in treffer[:5]
+            ],
+        ))
+    return ergebnis
+
 
 @router.get("", response_model=list[ProductOut])
 def list_products(
diff --git a/backend/app/routers/settings.py b/backend/app/routers/settings.py
index 418df99..3c4f53f 100644
--- a/backend/app/routers/settings.py
+++ b/backend/app/routers/settings.py
@@ -10,6 +10,10 @@ from ..schemas import SettingOut
 router = APIRouter(prefix="/settings", tags=["settings"])
 
 EXPIRY_WARNING_KEY = "expiry_warning_days"
+# Ab wie viel Prozent Übereinstimmung ein Artikel beim Kassenzettel-Scan als
+# Treffer vorgeschlagen wird.
+RECEIPT_THRESHOLD_KEY = "receipt_match_threshold"
+RECEIPT_THRESHOLD_DEFAULT = 45
 
 
 def get_expiry_warning_days(db: Session) -> int:
@@ -22,6 +26,17 @@ def get_expiry_warning_days(db: Session) -> int:
         return get_settings().expiry_warning_days_default
 
 
+def get_receipt_match_threshold(db: Session) -> int:
+    """Schwellwert (0–100) für Kassenzettel-Treffer; auf sinnvollen Bereich geklemmt."""
+    row = db.get(Setting, RECEIPT_THRESHOLD_KEY)
+    if row is None:
+        return RECEIPT_THRESHOLD_DEFAULT
+    try:
+        return max(0, min(100, int(row.value)))
+    except ValueError:
+        return RECEIPT_THRESHOLD_DEFAULT
+
+
 @router.get("", response_model=list[SettingOut])
 def list_settings(
     db: Session = Depends(get_db), _: User = Depends(get_current_user)
@@ -29,6 +44,7 @@ def list_settings(
     rows = db.query(Setting).all()
     known = {r.key: r.value for r in rows}
     known.setdefault(EXPIRY_WARNING_KEY, str(get_expiry_warning_days(db)))
+    known.setdefault(RECEIPT_THRESHOLD_KEY, str(get_receipt_match_threshold(db)))
     return [SettingOut(key=k, value=v) for k, v in known.items()]
 
 
diff --git a/backend/app/schemas.py b/backend/app/schemas.py
index 76881c8..de724c4 100644
--- a/backend/app/schemas.py
+++ b/backend/app/schemas.py
@@ -487,6 +487,25 @@ class LotSplit(BaseModel):
     location_id: str | None = None
 
 
+# ---- Kassenzettel-Abgleich ----
+class MatchRequest(BaseModel):
+    """Kassenzettel-Zeilen (OCR) → beste Lebensmittel-Treffer je Zeile."""
+    lines: list[str]
+    threshold: int | None = None   # 0–100; ohne Angabe gilt die Einstellung
+
+
+class MatchCandidate(BaseModel):
+    product_id: int
+    name: str
+    brand: str | None = None
+    score: int   # 0–100, Ähnlichkeit zur Kassenzeile
+
+
+class MatchLine(BaseModel):
+    text: str
+    candidates: list[MatchCandidate] = []
+
+
 class CheckInResponse(BaseModel):
     lot: LotOut
     product_stock: float
diff --git a/backend/tests/test_receipt_match.py b/backend/tests/test_receipt_match.py
new file mode 100644
index 0000000..695e101
--- /dev/null
+++ b/backend/tests/test_receipt_match.py
@@ -0,0 +1,60 @@
+"""Kassenzettel-Abgleich: OCR-Zeilen den ähnlichsten Lebensmitteln zuordnen."""
+
+import pytest
+
+from app.models import BaseUnit, Product, Role, User
+from app.routers.products import match_receipt
+from app.schemas import MatchRequest
+
+
+@pytest.fixture()
+def user(db):
+    person = User(username="tester", password_hash="x", role=Role.admin)
+    db.add(person)
+    db.commit()
+    db.refresh(person)
+    return person
+
+
+def test_abgekuerzte_kassenzeile_trifft_lebensmittel(db, user):
+    p = Product(
+        name="Vegane Mühlen-Schnitzel auf Basis von Soja",
+        base_unit=BaseUnit.gram, package_size=180,
+    )
+    db.add(p)
+    db.commit()
+    db.refresh(p)
+
+    res = match_receipt(
+        MatchRequest(lines=["MÜHLEN SCHNITZ", "voelliger unsinn xyz"]),
+        db=db, _=user,
+    )
+    # Abgekürzte Kassenzeile findet den vollen Namen mit hohem Score.
+    assert res[0].text == "MÜHLEN SCHNITZ"
+    assert res[0].candidates
+    assert res[0].candidates[0].product_id == p.id
+    assert res[0].candidates[0].score >= 70
+    # Unsinnszeile bleibt ohne Treffer über dem Schwellwert.
+    assert res[1].candidates == []
+
+
+def test_nur_lebensmittel_keine_gegenstaende(db, user):
+    # Einzelstück-Flag -> tracking "object" -> darf nicht vorgeschlagen werden.
+    obj = Product(name="Powerbank Anker", base_unit=BaseUnit.piece, individual=True)
+    db.add(obj)
+    db.commit()
+
+    res = match_receipt(MatchRequest(lines=["POWERBANK ANKER"], threshold=10), db=db, _=user)
+    assert res[0].candidates == []
+
+
+def test_schwellwert_filtert(db, user):
+    p = Product(name="Basmati Reis", base_unit=BaseUnit.gram, package_size=1)
+    db.add(p)
+    db.commit()
+
+    # Nur teilweise passende Zeile: bei sehr hohem Schwellwert kein Treffer.
+    hart = match_receipt(MatchRequest(lines=["reis lose"], threshold=99), db=db, _=user)
+    weich = match_receipt(MatchRequest(lines=["reis lose"], threshold=20), db=db, _=user)
+    assert hart[0].candidates == []
+    assert weich[0].candidates and weich[0].candidates[0].product_id == p.id