diff --git a/modules/video_coding/rtp_vp9_ref_finder.cc b/modules/video_coding/rtp_vp9_ref_finder.cc
index fd271f81f8..d4e30054ae 100644
--- a/modules/video_coding/rtp_vp9_ref_finder.cc
+++ b/modules/video_coding/rtp_vp9_ref_finder.cc
@@ -53,9 +53,6 @@ RtpVp9RefFinder::FrameDecision RtpVp9RefFinder::ManageFrameInternal(
   frame->SetSpatialIndex(codec_header.spatial_idx);
   frame->SetId(codec_header.picture_id & (kFrameIdLength - 1));
 
-  if (last_picture_id_ == -1)
-    last_picture_id_ = frame->Id();
-
   if (codec_header.flexible_mode) {
     if (codec_header.num_ref_pics > EncodedFrame::kMaxFrameReferences) {
       return kDrop;
diff --git a/modules/video_coding/rtp_vp9_ref_finder.h b/modules/video_coding/rtp_vp9_ref_finder.h
index 436cb1c84a..00de8ca59d 100644
--- a/modules/video_coding/rtp_vp9_ref_finder.h
+++ b/modules/video_coding/rtp_vp9_ref_finder.h
@@ -60,10 +60,6 @@ class RtpVp9RefFinder {
 
   void FlattenFrameIdAndRefs(RtpFrameObject* frame, bool inter_layer_predicted);
 
-  // Save the last picture id in order to detect when there is a gap in frames
-  // that have not yet been fully received.
-  int last_picture_id_ = -1;
-
   // Frames that have been fully received but didn't have all the information
   // needed to determine their references.
   std::deque<std::unique_ptr<RtpFrameObject>> stashed_frames_;