diff --git a/src/simplefractions/__init__.py b/src/simplefractions/__init__.py
index 15799f0..d1a84ea 100644
--- a/src/simplefractions/__init__.py
+++ b/src/simplefractions/__init__.py
@@ -37,7 +37,6 @@
 import fractions
 import math
 import numbers
-import struct
 import typing
 
 from simplefractions._simplest_in_interval import _simplest_in_interval
@@ -115,6 +114,11 @@ def _interval_rounding_to(
     """
     Return the interval of numbers that round to a given float.
 
+    Parameters
+    ----------
+    x
+        A finite float.
+
     Returns
     -------
     left, right : fractions.Fraction
@@ -123,22 +127,18 @@ def _interval_rounding_to(
     closed : bool
         True if the interval is closed at both ends, else False.
     """
-    if x < 0:
+    if x < 0.0:
         left, right, closed = _interval_rounding_to(-x)
         return -right, -left, closed
 
-    if x == 0:
-        n = struct.unpack("<Q", struct.pack("<d", 0.0))[0]
-        x_plus = struct.unpack("<d", struct.pack("<Q", n + 1))[0]
-        right = (fractions.Fraction(x) + fractions.Fraction(x_plus)) / 2
+    if x == 0.0:
+        right = fractions.Fraction(math.nextafter(0.0, math.inf)) / 2
         return -right, right, True
 
-    n = struct.unpack("<Q", struct.pack("<d", x))[0]
-    x_plus = struct.unpack("<d", struct.pack("<Q", n + 1))[0]
-    x_minus = struct.unpack("<d", struct.pack("<Q", n - 1))[0]
-
-    closed = n % 2 == 0
+    x_plus = math.nextafter(x, math.inf)
+    x_minus = math.nextafter(x, 0.0)
     left = (fractions.Fraction(x) + fractions.Fraction(x_minus)) / 2
+    closed = float(left) == x
     if math.isinf(x_plus):
         # Corner case where x was the largest representable finite float
         right = 2 * fractions.Fraction(x) - left
diff --git a/src/simplefractions/test/test_simplefractions.py b/src/simplefractions/test/test_simplefractions.py
index 1fdf22b..f66145f 100644
--- a/src/simplefractions/test/test_simplefractions.py
+++ b/src/simplefractions/test/test_simplefractions.py
@@ -244,6 +244,30 @@ def test_simplest_from_float(self) -> None:
             with self.subTest(f=f):
                 self.check_simplest_from_float(f)
 
+    def test_simplest_from_float_open_closed_cases(self) -> None:
+        # Cases where it matters whether the interval rounding to
+        # the float is open or closed.
+
+        # Even floats
+        self.assertEqual(simplest_from_float(1e16), 10**16 - 1)
+        self.assertEqual(simplest_from_float(1e16 + 4), 10**16 + 3)
+        self.assertEqual(simplest_from_float(float(2**54 - 4)), 2**54 - 5)
+        self.assertEqual(simplest_from_float(float(2**54)), 2**54 - 1)
+        self.assertEqual(simplest_from_float(float(2**54 + 8)), 2**54 + 6)
+
+        # Odd floats
+        self.assertEqual(simplest_from_float(1e16 + 2), 10**16 + 2)
+        self.assertEqual(simplest_from_float(float(2**54 - 2)), 2**54 - 2)
+        self.assertEqual(simplest_from_float(float(2**54 + 4)), 2**54 + 3)
+
+    def test_simplest_from_float_powers_of_two(self) -> None:
+        # Powers of two that are exactly representable in IEEE 754 binary64.
+        TWO = fractions.Fraction(2)
+        for e in range(-1074, 1024):
+            f = TWO**e
+            with self.subTest(f=f):
+                self.check_simplest_from_float(f)
+
     def test_simplest_from_float_special_values(self) -> None:
         with self.assertRaises(ValueError):
             simplest_from_float(math.inf)