diff --git a/lucene/CHANGES.txt b/lucene/CHANGES.txt
index 979f742b3082..1821a3bb55d7 100644
--- a/lucene/CHANGES.txt
+++ b/lucene/CHANGES.txt
@@ -368,6 +368,8 @@ Optimizations
 * GITHUB#16380: Speed up filtered disjunction queries when the filter is part of the
   index sort by advancing the filter before calculating impacts. (Alan Woodward)
 
+* GITHUB#16399: Reuse the previous end offset to skip a redundant addresses.get() in sorted numeric rangeIntoBitSet. (Sagar Upadhyaya)
+
 Bug Fixes
 ---------------------
 * GITHUB#16350: Disable bulk-scoring in monitor queries. (Alan Woodward)
diff --git a/lucene/core/src/java/org/apache/lucene/codecs/lucene90/Lucene90DocValuesProducer.java b/lucene/core/src/java/org/apache/lucene/codecs/lucene90/Lucene90DocValuesProducer.java
index 25eefc2660c7..83bfb934439f 100644
--- a/lucene/core/src/java/org/apache/lucene/codecs/lucene90/Lucene90DocValuesProducer.java
+++ b/lucene/core/src/java/org/apache/lucene/codecs/lucene90/Lucene90DocValuesProducer.java
@@ -2041,12 +2041,13 @@ public void rangeIntoBitSet(
                 v, fromDoc, endDoc, cardinality, lo, hi, bitSet, offset);
             return;
           }
+          long startOffset = addresses.get(fromDoc);
           for (int currentDoc = fromDoc; currentDoc < endDoc; currentDoc++) {
-            long startOffset = addresses.get(currentDoc);
             long endOffset = addresses.get(currentDoc + 1L);
             if (sortedNumericMatchesRange(v, startOffset, endOffset, lo, hi)) {
               bitSet.set(currentDoc - offset);
             }
+            startOffset = endOffset;
           }
         }