diff --git a/improver/categorical/decision_tree.py b/improver/categorical/decision_tree.py
index ed8299a71c..47e149124c 100644
--- a/improver/categorical/decision_tree.py
+++ b/improver/categorical/decision_tree.py
@@ -6,6 +6,7 @@
 
 import copy
 import operator
+import warnings
 from typing import Dict, List, Optional, Tuple, Union
 
 import iris
@@ -179,11 +180,13 @@ def prepare_input_cubes(
               indicated as allowed by the if_diagnostic_missing key.
 
         Raises:
-            ValueError:
-                Raises a ValueError if multiple matching thresholds are found.
             IOError:
                 Raises an IOError if any of the required input data is missing.
                 The error includes details of which fields are missing.
+
+        Warnings:
+            - If multiple thresholds have been found and the closest matching threshold
+            will be used.
         """
         cubes = as_cubelist(*cubes)
 
@@ -276,14 +279,33 @@ def prepare_input_cubes(
                             f"spp__relative_to_threshold: {condition}\n"
                         )
                     else:
-                        num_thresholds = len(
+                        threshold_points = (
                             matched_threshold[0].coord(threshold_name).points
                         )
+                        num_thresholds = len(threshold_points)
+
+                        # If multiple thresholds are found, the closest threshold to the
+                        # desired threshold, is chosen.
                         if num_thresholds > 1:
-                            raise ValueError(
+                            diff = [
+                                abs(point - threshold) for point in threshold_points
+                            ]
+                            closest_point = threshold_points[diff.index(min(diff))]
+
+                            warnings.warn(
                                 f"Multiple ({num_thresholds}) matching thresholds found"
-                                f" for name: {diagnostic}, threshold {threshold}"
+                                f" for name: {diagnostic}, threshold {threshold}. "
+                                f"Using closest match: {closest_point}"
                             )
+
+                            closest_point_constraint = iris.Constraint(
+                                coord_values={threshold_name: closest_point}
+                            )
+                            closest_cube = matched_threshold[0].extract(
+                                closest_point_constraint
+                            )
+                            used_cubes.append(closest_cube)
+
                         else:
                             used_cubes.extend(matched_threshold)
 
diff --git a/improver_tests/categorical/decision_tree/test_ApplyDecisionTree.py b/improver_tests/categorical/decision_tree/test_ApplyDecisionTree.py
index 715d72c4e6..cd4c4c5883 100644
--- a/improver_tests/categorical/decision_tree/test_ApplyDecisionTree.py
+++ b/improver_tests/categorical/decision_tree/test_ApplyDecisionTree.py
@@ -588,28 +588,39 @@ def test_returns_used_cubes(self):
         for constraint in unexpected:
             self.assertEqual(len(result.extract(constraint)), 0)
 
-    def test_raises_error_matching_threshold(self):
-        """Test prepare_input_cubes method raises error for matching thresholds in a
-        diagnostic."""
+    def test_selects_closest_threshold_when_multiple_matches(self):
+        """Test prepare_input_cubes method raises a warning for matching thresholds
+        in a diagnostic and selects the closest threshold."""
+
+        # Modify the cube snowfall_rate to add an additional threshold
         threshold_coord = find_threshold_coordinate(self.cubes[0])
-        additional_threshold = threshold_coord.points[0] * (
+        original_threshold = threshold_coord.points[0]
+        additional_threshold = original_threshold * (
             1 + 0.5 * self.plugin.float_tolerance
         )
         threshold_coord.points = np.array(
             [
-                threshold_coord.points[0],
+                original_threshold,
                 additional_threshold,
                 threshold_coord.points[2],
             ],
             dtype=np.float32,
         )
-        msg = (
-            r"Multiple \(2\) matching thresholds found for name: "
-            "probability_of_lwe_snowfall_rate_above_threshold"
-        )
-
-        with self.assertRaisesRegex(ValueError, msg):
-            self.plugin.prepare_input_cubes(self.cubes)
+        # Check warning is raised
+        msg = r"Multiple \(2\) matching thresholds found.*Using closest match"
+        with self.assertWarnsRegex(UserWarning, msg):
+            result, _ = self.plugin.prepare_input_cubes(self.cubes)
+        # Check thresholds extracted for snowfall_cubes
+        # We expect the original_threshold to be selected over the additional_threshold
+        expected_thresholds = {original_threshold, threshold_coord.points[2]}
+        snowfall_cubes = [cube for cube in result if "lwe_snowfall_rate" in cube.name()]
+        self.assertGreater(len(snowfall_cubes), 0)
+        for cube in snowfall_cubes:
+            cube_thresholds = cube.coord(find_threshold_coordinate(cube).name()).points
+            self.assertEqual(len(cube_thresholds), 1)
+            cube_threshold_value = cube_thresholds[0]
+            self.assertIn(cube_threshold_value, expected_thresholds)
+            self.assertNotEqual(cube_threshold_value, additional_threshold)
 
     def test_zero_threshold_uses_absolute_tolerance(self):
         """Test prepare_input_cubes method uses absolute tolerance when the threshold