diff --git a/Ska/Sun/data/pitch_roll.fits.gz b/Ska/Sun/data/pitch_roll.fits.gz
new file mode 100644
index 0000000..f74bc93
Binary files /dev/null and b/Ska/Sun/data/pitch_roll.fits.gz differ
diff --git a/Ska/Sun/sun.py b/Ska/Sun/sun.py
index 468f05f..9d23efc 100755
--- a/Ska/Sun/sun.py
+++ b/Ska/Sun/sun.py
@@ -2,15 +2,46 @@
 """
 Utility for calculating sun position and pitch angle.
 """
+from pathlib import Path
+import numpy as np
+from math import cos, sin, acos, atan2, asin, pi, radians, degrees
+from astropy.table import Table
 
+from ska_helpers.utils import LazyVal
 from Quaternion import Quat
 from chandra_aca.transform import radec_to_eci
 from Chandra.Time import DateTime
-from math import cos, sin, acos, atan2, asin, pi, radians, degrees
-import numpy as np
 import Ska.quatutil
 
 
+def load_roll_table():
+    dat = Table.read(Path(__file__).parent / 'data' / 'pitch_roll.fits.gz')
+
+    # Add a terminating row to the data such that for pitch at or greater
+    # than 180 the allowed roll deviation is defined as 0.
+    dat.add_row({'pitch': 180, 'rolldev': 0})
+    return dat
+
+
+ROLL_TABLE = LazyVal(load_roll_table)
+
+
+def allowed_rolldev(pitch):
+    """Get allowed roll deviation (off-nominal roll) for the given ``pitch``.
+    :param pitch: Sun pitch angle (deg)
+    :returns: Roll deviation (deg)
+    """
+    idx1 = np.searchsorted(ROLL_TABLE.val['pitch'], pitch, side='right')
+    idx0 = idx1 - 1
+    idx_max = len(ROLL_TABLE.val) - 1
+    idx0 = np.clip(idx0, 0, idx_max)
+    idx1 = np.clip(idx1, 0, idx_max)
+    val0 = ROLL_TABLE.val['rolldev'][idx0]
+    val1 = ROLL_TABLE.val['rolldev'][idx1]
+    out = np.minimum(val0, val1)  # works even for a vector input for `pitch`
+    return out
+
+
 # The position() method is a modification of
 # http://idlastro.gsfc.nasa.gov/ftp/pro/astro/sunpos.pro
 #
diff --git a/Ska/Sun/tests/__init__.py b/Ska/Sun/tests/__init__.py
new file mode 100644
index 0000000..e69de29
diff --git a/Ska/Sun/tests/test_sun.py b/Ska/Sun/tests/test_sun.py
index d11c0b7..4c420fe 100644
--- a/Ska/Sun/tests/test_sun.py
+++ b/Ska/Sun/tests/test_sun.py
@@ -3,7 +3,31 @@
 import numpy as np
 from Quaternion import Quat
 from ..sun import (apply_sun_pitch_yaw, get_sun_pitch_yaw, nominal_roll,
-                   off_nominal_roll, position, pitch as sun_pitch)
+                   off_nominal_roll, position, allowed_rolldev,
+                   pitch as sun_pitch)
+
+
+def test_allowed_rolldev():
+
+    # Test array of pitchs and allowed roll dev
+    testarr = [[135, 13.979],
+               [138, 14.516],
+               [0, 0],
+               [40, 0],
+               [179.9, 18.748772],
+               [179.997, 0],
+               [180, 0],
+               [181, 0],
+               [85.49229, 13.677669],
+               [85.52, 18.756727],
+               [124.99, 18.748772],
+               [125, 17.0]]
+    for pitch, rolldev in testarr:
+        assert np.isclose(allowed_rolldev(pitch), rolldev)
+
+    # Also test with pitch as vector
+    assert np.allclose(allowed_rolldev(np.array(testarr)[:, 0]),
+                       np.array(testarr)[:, 1])
 
 
 def test_position():
diff --git a/make_rolldev_file.py b/make_rolldev_file.py
new file mode 100644
index 0000000..f523ef9
--- /dev/null
+++ b/make_rolldev_file.py
@@ -0,0 +1,20 @@
+# For the pitch/roll table.  I used google sheets to save the xlsx file as
+# a CSV (after adding headers: pitch, negrolldev, rolldev).
+# For that CSV file, this little script checked for matching neg and
+# positive roll values and then saved out a compressed file.
+
+# For future versions of the file, we should convert directly
+# with something like
+# df = pandas.read_excel('pitch_roll_2022.xlsx'
+#    ...: , names=['pitch', 'rolldev_minus', 'rolldev_plus'])
+# dat = Table.from_pandas(df)
+
+import numpy as np
+from astropy.table import Table
+dat = Table.read('pitch_roll_2022.csv')
+
+assert np.allclose(np.abs(dat['negrolldev']), dat['rolldev'], rtol=0, atol=1e-08)
+
+dat['pitch'] = dat['pitch'].astype(np.float32)
+dat['rolldev'] = dat['rolldev'].astype(np.float32)
+dat['pitch', 'rolldev'].write('pitch_roll.fits.gz', overwrite=True)
diff --git a/setup.py b/setup.py
index 58a1095..d274b18 100755
--- a/setup.py
+++ b/setup.py
@@ -14,7 +14,7 @@
       setup_requires=['setuptools_scm', 'setuptools_scm_git_archive'],
       zip_safe=False,
       packages=['Ska', 'Ska.Sun', 'Ska.Sun.tests'],
-      package_data={},
+      package_data={'Ska.Sun': ['data/*fits.gz']},
       tests_require=['pytest'],
       cmdclass=cmdclass,
       )