diff --git a/pyproject.toml b/pyproject.toml
index 541c1a2d0..66be67324 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -25,7 +25,7 @@ classifiers = [
     'Topic :: Scientific/Engineering :: Artificial Intelligence',
 ]
 dependencies = [
-    "numpy >=1.21.0",
+    "numpy>=1.21.0,<2.0.0",
     "cloudpickle >=1.2.0",
     "importlib-metadata >=4.8.0; python_version < '3.10'",
     "typing-extensions >=4.3.0",
diff --git a/tests/envs/test_env_implementation.py b/tests/envs/test_env_implementation.py
index e75982b86..9cb51d446 100644
--- a/tests/envs/test_env_implementation.py
+++ b/tests/envs/test_env_implementation.py
@@ -6,7 +6,7 @@
 import gymnasium as gym
 from gymnasium.envs.box2d import BipedalWalker, CarRacing
 from gymnasium.envs.box2d.lunar_lander import demo_heuristic_lander
-from gymnasium.envs.toy_text import TaxiEnv
+from gymnasium.envs.toy_text import BlackjackEnv, TaxiEnv
 from gymnasium.envs.toy_text.frozen_lake import generate_random_map
 
 
@@ -256,6 +256,26 @@ def test_invalid_customizable_resets(env_name: str, low_high: list):
         env.reset(options={"low": low, "high": high})
 
 
+def test_blackjack_edgecases():
+    # GH 1011
+    # test dealer and player have natural 21
+    env = BlackjackEnv()
+    env.player = [1, 10]
+    env.dealer = [1, 10]
+    _obs, reward, terminated, _, _info = env.step(0)
+    # no payout expected
+    assert terminated is True
+    assert reward == 0
+
+    # test dealer and player have sum 21 but after a hit
+    env.reset()
+    env.player = [2, 10, 9]
+    env.dealer = [2, 10, 4, 5]
+    _obs, reward, terminated, _, _info = env.step(0)
+    assert terminated is True
+    assert reward == 0
+
+
 def test_cartpole_vector_equiv():
     env = gym.make("CartPole-v1")
     envs = gym.make_vec("CartPole-v1", num_envs=1)