diff --git a/ibis/expr/tests/test_newrels.py b/ibis/expr/tests/test_newrels.py
index 77f168d4474b..ff2f0c704ec6 100644
--- a/ibis/expr/tests/test_newrels.py
+++ b/ibis/expr/tests/test_newrels.py
@@ -1861,3 +1861,32 @@ def test_analytic_dereference():
     assert expr.op().predicates == (
         ops.Equals(ops.WindowFunction(ops.RowNumber()), ops.Literal(5, dtype="int8")),
     )
+
+
+def test_drop_null_schema_change():
+    orig_schema = ibis.schema({"a": "int64", "b": "string", "c": "!float64"})
+    t = ibis.table(orig_schema)
+
+    expr = t.drop_null()
+    expected_schema = ibis.schema({"a": "!int64", "b": "!string", "c": "!float64"})
+    assert expr.schema() == expected_schema
+
+    expr = t.drop_null("a")
+    expected_schema = ibis.schema({"a": "!int64", "b": "string", "c": "!float64"})
+    assert expr.schema() == expected_schema
+
+    expr = t.drop_null(["a", "c"])
+    expected_schema = ibis.schema({"a": "!int64", "b": "string", "c": "!float64"})
+    assert expr.schema() == expected_schema
+
+    expr = t.drop_null(how="all")
+    expected_schema = orig_schema
+    assert expr.schema() == expected_schema
+
+    expr = t.drop_null("a", how="all")
+    expected_schema = orig_schema
+    assert expr.schema() == expected_schema
+
+    expr = t.drop_null(["a", "c"], how="all")
+    expected_schema = orig_schema
+    assert expr.schema() == expected_schema
diff --git a/ibis/expr/types/relations.py b/ibis/expr/types/relations.py
index ef8f4a4f5908..b5c97c4f79d0 100644
--- a/ibis/expr/types/relations.py
+++ b/ibis/expr/types/relations.py
@@ -3061,9 +3061,17 @@ def drop_null(
         │ 344 │
         └─────┘
         """
-        if subset is not None:
-            subset = tuple(self.bind(subset))
-        return ops.DropNull(self, how, subset).to_expr()
+        subset_columns = None if subset is not None else tuple(self.bind(subset))
+        result = ops.DropNull(self, how, subset_columns).to_expr()
+        if how == "any":
+            # We now know that all columns in `subset` are non-nullable
+            schema = self.schema()
+            subset_names = (
+                schema.names if subset is None else util.promote_tuple(subset)
+            )
+            new_types = {col: schema[col].copy(nullable=False) for col in subset_names}
+            result = result.cast(new_types)
+        return result
 
     def fill_null(self, replacements: ir.Scalar | Mapping[str, ir.Scalar], /) -> Table:
         """Fill null values in a table expression.