diff --git a/modules/pacing/task_queue_paced_sender.cc b/modules/pacing/task_queue_paced_sender.cc
index ebfc7ece9a..f6591235b3 100644
--- a/modules/pacing/task_queue_paced_sender.cc
+++ b/modules/pacing/task_queue_paced_sender.cc
@@ -29,21 +29,6 @@ constexpr const char* kSlackedTaskQueuePacedSenderFieldTrial =
 
 }  // namespace
 
-TaskQueuePacedSender::TaskQueuePacedSender(
-    Clock* clock,
-    PacingController::PacketSender* packet_sender,
-    RtcEventLog* event_log,
-    const WebRtcKeyValueConfig* field_trials,
-    TaskQueueFactory* task_queue_factory,
-    TimeDelta max_hold_back_window,
-    int max_hold_back_window_in_packets)
-    : TaskQueuePacedSender(clock,
-                           packet_sender,
-                           *field_trials,
-                           task_queue_factory,
-                           max_hold_back_window,
-                           max_hold_back_window_in_packets) {}
-
 TaskQueuePacedSender::TaskQueuePacedSender(
     Clock* clock,
     PacingController::PacketSender* packet_sender,
diff --git a/modules/pacing/task_queue_paced_sender.h b/modules/pacing/task_queue_paced_sender.h
index 7a1d5f2fd7..4df8aafbdb 100644
--- a/modules/pacing/task_queue_paced_sender.h
+++ b/modules/pacing/task_queue_paced_sender.h
@@ -40,17 +40,6 @@ class Clock;
 
 class TaskQueuePacedSender : public RtpPacketPacer, public RtpPacketSender {
  public:
-  // TODO(bugs.webrtc.org/13417): Remove when downstream usage is gone.
-  ABSL_DEPRECATED("Use the other version instead.")
-  TaskQueuePacedSender(
-      Clock* clock,
-      PacingController::PacketSender* packet_sender,
-      RtcEventLog* event_log,
-      const WebRtcKeyValueConfig* field_trials,
-      TaskQueueFactory* task_queue_factory,
-      TimeDelta max_hold_back_window = PacingController::kMinSleepTime,
-      int max_hold_back_window_in_packets = -1);
-
   // The `hold_back_window` parameter sets a lower bound on time to sleep if
   // there is currently a pacer queue and packets can't immediately be
   // processed. Increasing this reduces thread wakeups at the expense of higher