diff --git a/doc/source/whatsnew/v1.4.2.rst b/doc/source/whatsnew/v1.4.2.rst
index 06f1f406c3816..b1d44b1388fdc 100644
--- a/doc/source/whatsnew/v1.4.2.rst
+++ b/doc/source/whatsnew/v1.4.2.rst
@@ -30,7 +30,7 @@ Bug fixes
 ~~~~~~~~~
 - Fix some cases for subclasses that define their ``_constructor`` properties as general callables (:issue:`46018`)
 - Fixed "longtable" formatting in :meth:`.Styler.to_latex` when ``column_format`` is given in extended format (:issue:`46037`)
--
+
 
 .. ---------------------------------------------------------------------------
 
diff --git a/doc/source/whatsnew/v1.5.0.rst b/doc/source/whatsnew/v1.5.0.rst
index 8dac952874f89..6aa64380b0a90 100644
--- a/doc/source/whatsnew/v1.5.0.rst
+++ b/doc/source/whatsnew/v1.5.0.rst
@@ -362,6 +362,7 @@ Conversion
 - Bug in :meth:`Series.astype` and :meth:`DataFrame.astype` from floating dtype to unsigned integer dtype failing to raise in the presence of negative values (:issue:`45151`)
 - Bug in :func:`array` with ``FloatingDtype`` and values containing float-castable strings incorrectly raising (:issue:`45424`)
 - Bug when comparing string and datetime64ns objects causing ``OverflowError`` exception. (:issue:`45506`)
+- Bug in :meth:`format.py` the percentiles values are converted to integer dtype, which ignores the floating point error while multiplying. (:issue:`46362`)
 
 Strings
 ^^^^^^^
diff --git a/pandas/io/formats/format.py b/pandas/io/formats/format.py
index 609fc2a45aa21..c337b2922f36e 100644
--- a/pandas/io/formats/format.py
+++ b/pandas/io/formats/format.py
@@ -1713,10 +1713,10 @@ def format_percentiles(
 
     percentiles = 100 * percentiles
 
-    int_idx = np.isclose(percentiles.astype(int), percentiles)
+    int_idx = np.isclose(percentiles.round(), percentiles)
 
     if np.all(int_idx):
-        out = percentiles.astype(int).astype(str)
+        out = percentiles.round().astype(int).astype(str)
         return [i + "%" for i in out]
 
     unique_pcts = np.unique(percentiles)
@@ -1729,7 +1729,7 @@ def format_percentiles(
     ).astype(int)
     prec = max(1, prec)
     out = np.empty_like(percentiles, dtype=object)
-    out[int_idx] = percentiles[int_idx].astype(int).astype(str)
+    out[int_idx] = percentiles[int_idx].round().astype(int).astype(str)
 
     out[~int_idx] = percentiles[~int_idx].round(prec).astype(str)
     return [i + "%" for i in out]
diff --git a/pandas/tests/io/formats/test_format.py b/pandas/tests/io/formats/test_format.py
index adcaeba5cfd8d..a8e260eaf6e51 100644
--- a/pandas/tests/io/formats/test_format.py
+++ b/pandas/tests/io/formats/test_format.py
@@ -3262,6 +3262,11 @@ def test_format_percentiles():
     expected = ["0%", "50%", "2.0%", "50%", "66.67%", "99.99%"]
     assert result == expected
 
+    # Issue #46362
+    result = fmt.format_percentiles([0.281,0.57,0.58,0.29])
+    expected = ['28.1%', '57%', '58%', '29%']
+    assert result == expected
+
     msg = r"percentiles should all be in the interval \[0,1\]"
     with pytest.raises(ValueError, match=msg):
         fmt.format_percentiles([0.1, np.nan, 0.5])