diff --git a/pom.xml b/pom.xml
index 02b611d915e..95066d84d93 100644
--- a/pom.xml
+++ b/pom.xml
@@ -988,7 +988,6 @@
                                 <exclude>org/elasticsearch/bootstrap/Bootstrap.class</exclude>
                                 <exclude>org/elasticsearch/Version.class</exclude>
                                 <exclude>org/apache/lucene/queries/XTermsFilter.class</exclude>
-                                <exclude>org/elasticsearch/index/percolator/stats/ShardPercolateService$RamEstimator.class</exclude>
                                 <exclude>org/elasticsearch/index/merge/Merges.class</exclude>
                                 <!-- end excludes for valid system-out -->
                                 <!-- start excludes for Unsafe -->
diff --git a/src/main/java/org/elasticsearch/index/percolator/stats/PercolateStats.java b/src/main/java/org/elasticsearch/index/percolator/stats/PercolateStats.java
index 4ebdc8aa176..e75813772de 100644
--- a/src/main/java/org/elasticsearch/index/percolator/stats/PercolateStats.java
+++ b/src/main/java/org/elasticsearch/index/percolator/stats/PercolateStats.java
@@ -18,6 +18,7 @@
  */
 package org.elasticsearch.index.percolator.stats;
 
+import org.elasticsearch.Version;
 import org.elasticsearch.common.io.stream.StreamInput;
 import org.elasticsearch.common.io.stream.StreamOutput;
 import org.elasticsearch.common.io.stream.Streamable;
@@ -37,7 +38,7 @@ public class PercolateStats implements Streamable, ToXContent {
     private long percolateCount;
     private long percolateTimeInMillis;
     private long current;
-    private long memorySizeInBytes;
+    private long memorySizeInBytes = -1;
     private long numQueries;
 
     /**
@@ -90,7 +91,9 @@ public class PercolateStats implements Streamable, ToXContent {
     }
 
     /**
-     * @return The total size the loaded queries take in memory.
+     * @return Temporarily returns <code>-1</code>, but this used to return the total size the loaded queries take in
+     * memory, but this is disabled now because the size estimation was too expensive cpu wise. This will be enabled
+     * again when a cheaper size estimation can be found.
      */
     public long getMemorySizeInBytes() {
         return memorySizeInBytes;
@@ -124,7 +127,6 @@ public class PercolateStats implements Streamable, ToXContent {
         percolateCount += percolate.getCount();
         percolateTimeInMillis += percolate.getTimeInMillis();
         current += percolate.getCurrent();
-        memorySizeInBytes += percolate.getMemorySizeInBytes();
         numQueries += percolate.getNumQueries();
     }
 
@@ -150,7 +152,11 @@ public class PercolateStats implements Streamable, ToXContent {
         percolateCount = in.readVLong();
         percolateTimeInMillis = in.readVLong();
         current = in.readVLong();
-        memorySizeInBytes = in.readVLong();
+        if (in.getVersion().before(Version.V_1_1_0)) {
+            in.readVLong();
+        } else {
+            in.readLong();
+        }
         numQueries = in.readVLong();
     }
 
@@ -159,7 +165,11 @@ public class PercolateStats implements Streamable, ToXContent {
         out.writeVLong(percolateCount);
         out.writeVLong(percolateTimeInMillis);
         out.writeVLong(current);
-        out.writeVLong(memorySizeInBytes);
+        if (out.getVersion().before(Version.V_1_1_0)) {
+            out.writeVLong(0);
+        } else {
+            out.writeLong(-1);
+        }
         out.writeVLong(numQueries);
     }
 }
diff --git a/src/main/java/org/elasticsearch/index/percolator/stats/ShardPercolateService.java b/src/main/java/org/elasticsearch/index/percolator/stats/ShardPercolateService.java
index a6b30f0f4db..e60a37c779c 100644
--- a/src/main/java/org/elasticsearch/index/percolator/stats/ShardPercolateService.java
+++ b/src/main/java/org/elasticsearch/index/percolator/stats/ShardPercolateService.java
@@ -20,7 +20,6 @@
 package org.elasticsearch.index.percolator.stats;
 
 import org.apache.lucene.search.Query;
-import org.apache.lucene.util.RamUsageEstimator;
 import org.elasticsearch.common.inject.Inject;
 import org.elasticsearch.common.lucene.HashedBytesRef;
 import org.elasticsearch.common.metrics.CounterMetric;
@@ -38,7 +37,6 @@ import java.util.concurrent.TimeUnit;
  *     <li> total time spent in percolate api
  *     <li> the current number of percolate requests
  *     <li> number of registered percolate queries
- *     <li> the estimated amount of memory the registered queries take
  * </ul>
  */
 public class ShardPercolateService extends AbstractIndexShardComponent {
@@ -52,7 +50,6 @@ public class ShardPercolateService extends AbstractIndexShardComponent {
     private final CounterMetric currentMetric = new CounterMetric();
 
     private final CounterMetric numberOfQueries = new CounterMetric();
-    private final CounterMetric memorySizeInBytes = new CounterMetric();
 
     public void prePercolate() {
         currentMetric.inc();
@@ -64,27 +61,22 @@ public class ShardPercolateService extends AbstractIndexShardComponent {
     }
 
     public void addedQuery(HashedBytesRef id, Query previousQuery, Query newQuery) {
-        if (previousQuery != null) {
-            memorySizeInBytes.dec(computeSizeInMemory(id, previousQuery));
-        } else {
-            numberOfQueries.inc();
-        }
-        memorySizeInBytes.inc(computeSizeInMemory(id, newQuery));
+        numberOfQueries.inc();
     }
 
     public void removedQuery(HashedBytesRef id, Query query) {
         numberOfQueries.dec();
-        memorySizeInBytes.dec(computeSizeInMemory(id, query));
     }
 
     /**
      * @return The current metrics
      */
     public PercolateStats stats() {
-        return new PercolateStats(percolateMetric.count(), TimeUnit.NANOSECONDS.toMillis(percolateMetric.sum()), currentMetric.count(), memorySizeInBytes.count(), numberOfQueries.count());
+        return new PercolateStats(percolateMetric.count(), TimeUnit.NANOSECONDS.toMillis(percolateMetric.sum()), currentMetric.count(), -1, numberOfQueries.count());
     }
 
-    private static long computeSizeInMemory(HashedBytesRef id, Query query) {
+    // Enable when a more efficient manner is found for estimating the size of a Lucene query.
+    /*private static long computeSizeInMemory(HashedBytesRef id, Query query) {
         long size = (3 * RamUsageEstimator.NUM_BYTES_INT) + RamUsageEstimator.NUM_BYTES_OBJECT_REF + RamUsageEstimator.NUM_BYTES_OBJECT_HEADER + id.bytes.bytes.length;
         size += RamEstimator.sizeOf(query);
         return size;
@@ -96,6 +88,6 @@ public class ShardPercolateService extends AbstractIndexShardComponent {
         static long sizeOf(Query query) {
             return RamUsageEstimator.sizeOf(query);
         }
-    }
+    }*/
 
 }
diff --git a/src/test/java/org/elasticsearch/percolator/PercolatorTests.java b/src/test/java/org/elasticsearch/percolator/PercolatorTests.java
index 03977ca0994..24474ef7757 100644
--- a/src/test/java/org/elasticsearch/percolator/PercolatorTests.java
+++ b/src/test/java/org/elasticsearch/percolator/PercolatorTests.java
@@ -558,7 +558,7 @@ public class PercolatorTests extends ElasticsearchIntegrationTest {
         assertThat(indicesResponse.getTotal().getPercolate().getCount(), equalTo((long) numShards.numPrimaries));
         assertThat(indicesResponse.getTotal().getPercolate().getCurrent(), equalTo(0l));
         assertThat(indicesResponse.getTotal().getPercolate().getNumQueries(), equalTo((long)numShards.dataCopies)); //number of copies
-        assertThat(indicesResponse.getTotal().getPercolate().getMemorySizeInBytes(), greaterThan(0l));
+        assertThat(indicesResponse.getTotal().getPercolate().getMemorySizeInBytes(), equalTo(-1l));
 
         NodesStatsResponse nodesResponse = client().admin().cluster().prepareNodesStats().execute().actionGet();
         long percolateCount = 0;
@@ -580,7 +580,7 @@ public class PercolatorTests extends ElasticsearchIntegrationTest {
         assertThat(indicesResponse.getTotal().getPercolate().getCount(), equalTo((long) numShards.numPrimaries * 2));
         assertThat(indicesResponse.getTotal().getPercolate().getCurrent(), equalTo(0l));
         assertThat(indicesResponse.getTotal().getPercolate().getNumQueries(), equalTo((long)numShards.dataCopies)); //number of copies
-        assertThat(indicesResponse.getTotal().getPercolate().getMemorySizeInBytes(), greaterThan(0l));
+        assertThat(indicesResponse.getTotal().getPercolate().getMemorySizeInBytes(), equalTo(-1l));
 
         percolateCount = 0;
         nodesResponse = client().admin().cluster().prepareNodesStats().execute().actionGet();