diff --git a/jedi/inference/gradual/annotation.py b/jedi/inference/gradual/annotation.py
index eadcfc2e..81780010 100644
--- a/jedi/inference/gradual/annotation.py
+++ b/jedi/inference/gradual/annotation.py
@@ -196,13 +196,43 @@ def py__annotations__(funcdef):
     return dct
 
 
+def resolve_forward_references(context, all_annotations):
+    def resolve(node):
+        if node is None or node.type != 'string':
+            return node
+
+        node = _get_forward_reference_node(
+            context,
+            context.inference_state.compiled_subprocess.safe_literal_eval(
+                node.value,
+            ),
+        )
+
+        if node is None:
+            # There was a string, but it's not a valid annotation
+            return None
+
+        # The forward reference tree has an additional root node ('eval_input')
+        # that we don't want. Extract the node we do want, that is equivalent to
+        # the nodes returned by `py__annotations__` for a non-quoted node.
+        node = node.children[0]
+
+        return node
+
+    return {name: resolve(node) for name, node in all_annotations.items()}
+
+
 @inference_state_method_cache()
 def infer_return_types(function, arguments):
     """
     Infers the type of a function's return value,
     according to type annotations.
     """
-    all_annotations = py__annotations__(function.tree_node)
+    context = function.get_default_param_context()
+    all_annotations = resolve_forward_references(
+        context,
+        py__annotations__(function.tree_node),
+    )
     annotation = all_annotations.get("return", None)
     if annotation is None:
         # If there is no Python 3-type annotation, look for an annotation
@@ -217,11 +247,10 @@ def infer_return_types(function, arguments):
             return NO_VALUES
 
         return _infer_annotation_string(
-            function.get_default_param_context(),
+            context,
             match.group(1).strip()
         ).execute_annotation()
 
-    context = function.get_default_param_context()
     unknown_type_vars = find_unknown_type_vars(context, annotation)
     annotation_values = infer_annotation(context, annotation)
     if not unknown_type_vars:
diff --git a/test/completion/pep0484_generic_passthroughs.py b/test/completion/pep0484_generic_passthroughs.py
index 49b133a8..7c7b8820 100644
--- a/test/completion/pep0484_generic_passthroughs.py
+++ b/test/completion/pep0484_generic_passthroughs.py
@@ -58,6 +58,16 @@ def typed_bound_generic_passthrough(x: TList) -> TList:
 
     return x
 
+# Forward references are more likely with custom types, however this aims to
+# test just the handling of the quoted type rather than any other part of the
+# machinery.
+def typed_quoted_return_generic_passthrough(x: T) -> 'List[T]':
+    return [x]
+
+def typed_quoted_input_generic_passthrough(x: 'Tuple[T]') -> T:
+    x
+    return x[0]
+
 
 for a in untyped_passthrough(untyped_list_str):
     #? str()
@@ -146,6 +156,23 @@ for q in typed_bound_generic_passthrough(typed_list_str):
     q
 
 
+for r in typed_quoted_return_generic_passthrough("something"):
+    #? str()
+    r
+
+for s in typed_quoted_return_generic_passthrough(42):
+    #? int()
+    s
+
+
+#? str()
+typed_quoted_input_generic_passthrough(("something",))
+
+#? int()
+typed_quoted_input_generic_passthrough((42,))
+
+
+
 class CustomList(List):
     def get_first(self):
         return self[0]