diff --git a/pandas/core/internals/construction.py b/pandas/core/internals/construction.py
index 7f3d246a6fda6..1a85dbe70a5e3 100644
--- a/pandas/core/internals/construction.py
+++ b/pandas/core/internals/construction.py
@@ -456,6 +456,17 @@ def dict_to_mgr(
         ]
         # TODO: can we get rid of the dt64tz special case above?
 
+    if dtype is None and index is None:
+        data_len = 0
+        for each in data.values():
+            if isinstance(each, (int, str)):
+                data_len += 1
+            elif each is None:
+                continue
+            else:
+                data_len += len(each)
+        if data_len == 0:
+            dtype = object
     return arrays_to_mgr(
         arrays, data_names, index, columns, dtype=dtype, typ=typ, consolidate=copy
     )
diff --git a/pandas/tests/dtypes/test_dtypes.py b/pandas/tests/dtypes/test_dtypes.py
index abb29ce66fd34..414f81fdd5a2d 100644
--- a/pandas/tests/dtypes/test_dtypes.py
+++ b/pandas/tests/dtypes/test_dtypes.py
@@ -1095,3 +1095,10 @@ def test_period_dtype_compare_to_string():
     dtype = PeriodDtype(freq="M")
     assert (dtype == "period[M]") is True
     assert (dtype != "period[M]") is False
+
+
+def test_dataframe_constructor_dtype():
+    # GH#42971
+    expected = pd.DataFrame(columns=["a"])
+    result = pd.DataFrame({"a": []})
+    tm.assert_series_equal(result.dtypes, expected.dtypes)