diff --git a/doc/whats_new/v1.3.rst b/doc/whats_new/v1.3.rst
index 7ba9c8a119836..9cab0db995c5d 100644
--- a/doc/whats_new/v1.3.rst
+++ b/doc/whats_new/v1.3.rst
@@ -516,7 +516,7 @@ Changelog
   :class:`tree.DecisionTreeClassifier` support missing values when
   `splitter='best'` and criterion is `gini`, `entropy`, or `log_loss`,
   for classification or `squared_error`, `friedman_mse`, or `poisson`
-  for regression. :pr:`23595` by `Thomas Fan`_.
+  for regression. :pr:`23595`, :pr:`26376` by `Thomas Fan`_.
 
 - |Enhancement| Adds a `class_names` parameter to
   :func:`tree.export_text`. This allows specifying the parameter `class_names`
diff --git a/sklearn/tree/_criterion.pyx b/sklearn/tree/_criterion.pyx
index 2f8a99fe7a26e..ef3b3e119d3aa 100644
--- a/sklearn/tree/_criterion.pyx
+++ b/sklearn/tree/_criterion.pyx
@@ -838,7 +838,9 @@ cdef class RegressionCriterion(Criterion):
         self.sample_indices[-n_missing:]
         """
         cdef SIZE_t i, p, k
-        cdef DOUBLE_t w = 0.0
+        cdef DOUBLE_t y_ik
+        cdef DOUBLE_t w_y_ik
+        cdef DOUBLE_t w = 1.0
 
         self.n_missing = n_missing
         if n_missing == 0:
@@ -855,7 +857,9 @@ cdef class RegressionCriterion(Criterion):
                 w = self.sample_weight[i]
 
             for k in range(self.n_outputs):
-                self.sum_missing[k] += w
+                y_ik = self.y[i, k]
+                w_y_ik = w * y_ik
+                self.sum_missing[k] += w_y_ik
 
             self.weighted_n_missing += w
 
diff --git a/sklearn/tree/tests/test_tree.py b/sklearn/tree/tests/test_tree.py
index ea3e40fddb7a5..eefae6cdaa3f6 100644
--- a/sklearn/tree/tests/test_tree.py
+++ b/sklearn/tree/tests/test_tree.py
@@ -2549,7 +2549,8 @@ def test_missing_values_poisson():
         (datasets.make_classification, DecisionTreeClassifier),
     ],
 )
-def test_missing_values_is_resilience(make_data, Tree):
+@pytest.mark.parametrize("sample_weight_train", [None, "ones"])
+def test_missing_values_is_resilience(make_data, Tree, sample_weight_train):
     """Check that trees can deal with missing values and have decent performance."""
 
     rng = np.random.RandomState(0)
@@ -2563,15 +2564,18 @@ def test_missing_values_is_resilience(make_data, Tree):
         X_missing, y, random_state=0
     )
 
+    if sample_weight_train == "ones":
+        sample_weight_train = np.ones(X_missing_train.shape[0])
+
     # Train tree with missing values
     tree_with_missing = Tree(random_state=rng)
-    tree_with_missing.fit(X_missing_train, y_train)
+    tree_with_missing.fit(X_missing_train, y_train, sample_weight=sample_weight_train)
     score_with_missing = tree_with_missing.score(X_missing_test, y_test)
 
     # Train tree without missing values
     X_train, X_test, y_train, y_test = train_test_split(X, y, random_state=0)
     tree = Tree(random_state=rng)
-    tree.fit(X_train, y_train)
+    tree.fit(X_train, y_train, sample_weight=sample_weight_train)
     score_without_missing = tree.score(X_test, y_test)
 
     # Score is still 90 percent of the tree's score that had no missing values
@@ -2601,3 +2605,32 @@ def test_missing_value_is_predictive():
 
     assert tree.score(X_train, y_train) >= 0.85
     assert tree.score(X_test, y_test) >= 0.85
+
+
+@pytest.mark.parametrize(
+    "make_data, Tree",
+    [
+        (datasets.make_regression, DecisionTreeRegressor),
+        (datasets.make_classification, DecisionTreeClassifier),
+    ],
+)
+def test_sample_weight_non_uniform(make_data, Tree):
+    """Check sample weight is correctly handled with missing values."""
+    rng = np.random.RandomState(0)
+    n_samples, n_features = 1000, 10
+    X, y = make_data(n_samples=n_samples, n_features=n_features, random_state=rng)
+
+    # Create dataset with missing values
+    X[rng.choice([False, True], size=X.shape, p=[0.9, 0.1])] = np.nan
+
+    # Zero sample weight is the same as removing the sample
+    sample_weight = np.ones(X.shape[0])
+    sample_weight[::2] = 0.0
+
+    tree_with_sw = Tree(random_state=0)
+    tree_with_sw.fit(X, y, sample_weight=sample_weight)
+
+    tree_samples_removed = Tree(random_state=0)
+    tree_samples_removed.fit(X[1::2, :], y[1::2])
+
+    assert_allclose(tree_samples_removed.predict(X), tree_with_sw.predict(X))