diff --git a/formulaic/materializers/base.py b/formulaic/materializers/base.py
index a717f771..9a097eb5 100644
--- a/formulaic/materializers/base.py
+++ b/formulaic/materializers/base.py
@@ -50,7 +50,6 @@
 if TYPE_CHECKING:  # pragma: no cover
     from formulaic import FormulaSpec, ModelSpec, ModelSpecs
 
-
 EncodedTermStructure = namedtuple(
     "EncodedTermStructure", ("term", "scoped_terms", "columns")
 )
@@ -590,7 +589,12 @@ def _evaluate_factor(
         return self.factor_cache[factor.expr]
 
     def _lookup(self, name: str) -> Tuple[Any, Set[Variable]]:
-        values, layer = self.layered_context.get_with_layer_name(name)
+        sentinel = object()
+        values, layer = self.layered_context.get_with_layer_name(name, default=sentinel)
+        if values is sentinel:
+            raise NameError(
+                f"`{name}` is not present in the dataset or evaluation context."
+            )
         return values, {Variable(name, roles=("value",), source=layer)}
 
     def _evaluate(
diff --git a/tests/materializers/test_pandas.py b/tests/materializers/test_pandas.py
index 2c6de395..889616e8 100644
--- a/tests/materializers/test_pandas.py
+++ b/tests/materializers/test_pandas.py
@@ -455,3 +455,20 @@ def test_quoted_python_args(self):
         assert mm.shape == (3, 2)
         assert len(mm.model_spec.structure) == 2
         assert numpy.all(mm.values == numpy.array([[1, 1], [1, 4], [1, 9]]))
+
+    def test_lookup_nonexistent_variable(self):
+        data = pandas.DataFrame({})
+        with pytest.raises(
+            FactorEvaluationError,
+            match=re.escape(
+                "Unable to evaluate factor `a`. [NameError: `a` is not present in the dataset or evaluation context.]"
+            ),
+        ):
+            PandasMaterializer(data).get_model_matrix("a")
+        with pytest.raises(
+            FactorEvaluationError,
+            match=re.escape(
+                "Unable to evaluate factor `I(a)`. [NameError: name 'a' is not defined]"
+            ),
+        ):
+            PandasMaterializer(data).get_model_matrix("I(a)")