diff --git a/.github/workflows/test.yml b/.github/workflows/test.yml
index 0672c84..025418a 100644
--- a/.github/workflows/test.yml
+++ b/.github/workflows/test.yml
@@ -34,6 +34,10 @@ jobs:
       matrix:
         python-version: ["3.10", "3.11", "3.12", "3.13", "3.14"]
         os: [ubuntu-latest, windows-latest]
+        pandas-version: [pd2, pd3]
+        exclude:
+          - python-version: "3.10" # pd3 requires python >= 3.11
+            pandas-version: pd3
 
     steps:
       - uses: actions/checkout@v6
@@ -57,4 +61,4 @@ jobs:
       - name: Test
         shell: bash
         run: |
-          tox run -e py --installpkg `find dist/*.tar.gz`
+          tox run -e ${{ matrix.pandas-version }} --installpkg `find dist/*.tar.gz`
diff --git a/CHANGELOG.rst b/CHANGELOG.rst
index c0ca09e..615fdd9 100644
--- a/CHANGELOG.rst
+++ b/CHANGELOG.rst
@@ -1,3 +1,10 @@
+UNRELEASED
+----------
+
+*UNRELEASED*
+
+* `#225 <https://github.com/ESSS/pytest-regressions/issues/225>`__: Add support for Pandas 3.0 to `dataframe_regression`.
+
 2.9.1
 -----
 
diff --git a/src/pytest_regressions/dataframe_regression.py b/src/pytest_regressions/dataframe_regression.py
index 034846c..eaa6153 100644
--- a/src/pytest_regressions/dataframe_regression.py
+++ b/src/pytest_regressions/dataframe_regression.py
@@ -59,8 +59,8 @@ def _check_data_types(
             raise ModuleNotFoundError(import_error_message("NumPy"))
 
         __tracebackhide__ = True
-        obtained_data_type = obtained_column.values.dtype
-        expected_data_type = expected_column.values.dtype
+        obtained_data_type = obtained_column.values.dtype.type
+        expected_data_type = expected_column.values.dtype.type
         if obtained_data_type != expected_data_type:
             # Check if both data types are comparable as numbers (float, int, short, bytes, etc...)
             if np.issubdtype(obtained_data_type, np.number) and np.issubdtype(
@@ -134,7 +134,7 @@ def _check_fn(self, obtained_filename: Path, expected_filename: Path) -> None:
             self._check_data_types(k, obtained_column, expected_column)
             self._check_data_shapes(obtained_column, expected_column)
 
-            if np.issubdtype(obtained_column.values.dtype, np.inexact):
+            if np.issubdtype(obtained_column.values.dtype.type, np.inexact):
                 not_close_mask = ~np.isclose(
                     obtained_column.values,
                     expected_column.values,
@@ -156,7 +156,10 @@ def _check_fn(self, obtained_filename: Path, expected_filename: Path) -> None:
                 diff_expected_data = expected_column[diff_ids]
                 if obtained_column.values.dtype == bool:
                     diffs = np.logical_xor(obtained_column, expected_column)[diff_ids]
-                elif obtained_column.values.dtype == object:
+                elif (
+                    obtained_column.values.dtype == object
+                    or obtained_column.values.dtype == "str"
+                ):
                     diffs = diff_obtained_data.copy()
                     diffs[:] = "?"
                 else:
@@ -173,7 +176,10 @@ def _check_fn(self, obtained_filename: Path, expected_filename: Path) -> None:
             error_msg += "To update values, use --force-regen option.\n\n"
             for k, comparison_table in comparison_tables_dict.items():
                 error_msg += f"{k}:\n{comparison_table}\n\n"
-            if obtained_column.values.dtype == object:
+            if (
+                obtained_column.values.dtype == object
+                or obtained_column.values.dtype == "str"
+            ):
                 error_msg += (
                     "WARNING: diffs for this kind of data type cannot be computed."
                 )
@@ -255,7 +261,7 @@ def check(
                 continue
 
             # Arrays of strings are supported.
-            if (array.dtype == "O") and (type(array.iloc[0]) is str):
+            if (array.dtype.kind == "O") and (type(array.iloc[0]) is str):
                 continue
             # Rejected: timedelta, objects, zero-terminated bytes, unicode strings and raw data
             assert array.dtype.kind not in ["m", "O", "S", "U", "V"], (
diff --git a/tox.ini b/tox.ini
index 2cc2c5a..08d0819 100644
--- a/tox.ini
+++ b/tox.ini
@@ -1,5 +1,5 @@
 [tox]
-envlist = py310,py311,py312,py313,py314,pytest6
+envlist = py310,py311,py312,py313,py314,pytest6,pd2,pd3
 
 [testenv]
 download = true
@@ -7,6 +7,8 @@ extras = dev
 commands = pytest {posargs:tests}
 deps =
     pytest6: pytest ~= 6.2
+    pd2: pandas < 3.0.0
+    pd3: pandas >= 3.0.0
 
 
 [pytest]