diff --git a/pandas/core/reshape/encoding.py b/pandas/core/reshape/encoding.py
index ad4a5db441b89..67fb075110f0d 100644
--- a/pandas/core/reshape/encoding.py
+++ b/pandas/core/reshape/encoding.py
@@ -390,7 +390,9 @@ def from_dummies(
         The default category is the implied category when a value has none of the
         listed categories specified with a one, i.e. if all dummies in a row are
         zero. Can be a single value for all variables or a dict directly mapping
-        the default categories to a prefix of a variable.
+        the default categories to a prefix of a variable. The default category
+        will be coerced to the dtype of ``data.columns`` if such coercion is
+        lossless, and will raise otherwise.
 
     Returns
     -------
diff --git a/pandas/tests/reshape/test_from_dummies.py b/pandas/tests/reshape/test_from_dummies.py
index c7b7992a78232..dfb691c785404 100644
--- a/pandas/tests/reshape/test_from_dummies.py
+++ b/pandas/tests/reshape/test_from_dummies.py
@@ -333,9 +333,7 @@ def test_no_prefix_string_cats_default_category(
 ):
     dummies = DataFrame({"a": [1, 0, 0], "b": [0, 1, 0]})
     result = from_dummies(dummies, default_category=default_category)
-    expected = DataFrame(expected)
-    if using_infer_string:
-        expected[""] = expected[""].astype("str")
+    expected = DataFrame(expected, dtype=dummies.columns.dtype)
     tm.assert_frame_equal(result, expected)
 
 
@@ -449,3 +447,31 @@ def test_maintain_original_index():
     result = from_dummies(df)
     expected = DataFrame({"": list("abca")}, index=list("abcd"))
     tm.assert_frame_equal(result, expected)
+
+
+def test_int_columns_with_float_default():
+    # https://github.com/pandas-dev/pandas/pull/60694
+    df = DataFrame(
+        {
+            3: [1, 0, 0],
+            4: [0, 1, 0],
+        },
+    )
+    with pytest.raises(ValueError, match="Trying to coerce float values to integers"):
+        from_dummies(df, default_category=0.5)
+
+
+def test_object_dtype_preserved():
+    # https://github.com/pandas-dev/pandas/pull/60694
+    # When the input has object dtype, the result should as
+    # well even when infer_string is True.
+    df = DataFrame(
+        {
+            "x": [1, 0, 0],
+            "y": [0, 1, 0],
+        },
+    )
+    df.columns = df.columns.astype("object")
+    result = from_dummies(df, default_category="z")
+    expected = DataFrame({"": ["x", "y", "z"]}, dtype="object")
+    tm.assert_frame_equal(result, expected)