diff --git a/sys/net/bpf.c b/sys/net/bpf.c
--- a/sys/net/bpf.c
+++ b/sys/net/bpf.c
@@ -261,6 +261,8 @@
 	.f_event = filt_bpfwrite,
 };
 
+static eventhandler_tag	bpf_ifdetach_cookie = NULL;
+
 /*
  * LOCKING MODEL USED BY BPF
  *
@@ -2887,6 +2889,39 @@
 	BPF_UNLOCK();
 }
 
+/*
+ * Interface departure handler. Detach descriptors on departure event.
+ * Note departure event does not guarantee interface is going down.
+ * Interface renaming is currently done via departure/arrival event set.
+ */
+static void
+bpf_ifdetach(void *arg __unused, struct ifnet *ifp)
+{
+	struct bpf_if *bp;
+	struct bpf_d *d;
+
+	/* Ignore ifnet renaming. */
+	if ((ifp->if_flags & IFF_RENAMING) != 0)
+		return;
+
+	BPF_LOCK();
+	CK_LIST_FOREACH(bp, &bpf_iflist, bif_next) {
+		if (bp->bif_ifp != ifp)
+			continue;
+
+		/* Detach common descriptors */
+		while ((d = CK_LIST_FIRST(&bp->bif_dlist)) != NULL) {
+			bpf_detachd_locked(d, true);
+		}
+
+		/* Detach writer-only descriptors */
+		while ((d = CK_LIST_FIRST(&bp->bif_wlist)) != NULL) {
+			bpf_detachd_locked(d, true);
+		}
+	}
+	BPF_UNLOCK();
+}
+
 bool
 bpf_peers_present_if(struct ifnet *ifp)
 {
@@ -2984,6 +3019,11 @@
 	dev = make_dev(&bpf_cdevsw, 0, UID_ROOT, GID_WHEEL, 0600, "bpf");
 	/* For compatibility */
 	make_dev_alias(dev, "bpf0");
+
+	/* Register interface departure handler */
+	bpf_ifdetach_cookie = EVENTHANDLER_REGISTER(
+		ifnet_departure_event, bpf_ifdetach, NULL,
+		EVENTHANDLER_PRI_ANY);
 }
 
 /*