diff --git a/tools/perf/util/event.c b/tools/perf/util/event.c
index 1f08f008d289..2fbf6a463c81 100644
--- a/tools/perf/util/event.c
+++ b/tools/perf/util/event.c
@@ -538,8 +538,10 @@ int event__process_task(event_t *self, struct perf_session *session)
 	dump_printf("(%d:%d):(%d:%d)\n", self->fork.pid, self->fork.tid,
 		    self->fork.ppid, self->fork.ptid);
 
-	if (self->header.type == PERF_RECORD_EXIT)
+	if (self->header.type == PERF_RECORD_EXIT) {
+		perf_session__remove_thread(session, thread);
 		return 0;
+	}
 
 	if (thread == NULL || parent == NULL ||
 	    thread__fork(thread, parent) < 0) {
diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c
index 8f83a1835766..c422cd676313 100644
--- a/tools/perf/util/session.c
+++ b/tools/perf/util/session.c
@@ -90,6 +90,7 @@ struct perf_session *perf_session__new(const char *filename, int mode, bool forc
 
 	memcpy(self->filename, filename, len);
 	self->threads = RB_ROOT;
+	INIT_LIST_HEAD(&self->dead_threads);
 	self->hists_tree = RB_ROOT;
 	self->last_match = NULL;
 	self->mmap_window = 32;
@@ -131,6 +132,16 @@ void perf_session__delete(struct perf_session *self)
 	free(self);
 }
 
+void perf_session__remove_thread(struct perf_session *self, struct thread *th)
+{
+	rb_erase(&th->rb_node, &self->threads);
+	/*
+	 * We may have references to this thread, for instance in some hist_entry
+	 * instances, so just move them to a separate list.
+	 */
+	list_add_tail(&th->node, &self->dead_threads);
+}
+
 static bool symbol__match_parent_regex(struct symbol *sym)
 {
 	if (sym->name && !regexec(&parent_regex, sym->name, 0, NULL, 0))
diff --git a/tools/perf/util/session.h b/tools/perf/util/session.h
index 55c6881b218d..9fa0fc2a863f 100644
--- a/tools/perf/util/session.h
+++ b/tools/perf/util/session.h
@@ -26,6 +26,7 @@ struct perf_session {
 	unsigned long		size;
 	unsigned long		mmap_window;
 	struct rb_root		threads;
+	struct list_head	dead_threads;
 	struct thread		*last_match;
 	struct machine		host_machine;
 	struct rb_root		machines;
@@ -99,6 +100,7 @@ int perf_session__create_kernel_maps(struct perf_session *self);
 
 int do_read(int fd, void *buf, size_t size);
 void perf_session__update_sample_type(struct perf_session *self);
+void perf_session__remove_thread(struct perf_session *self, struct thread *th);
 
 static inline
 struct machine *perf_session__find_host_machine(struct perf_session *self)
diff --git a/tools/perf/util/thread.h b/tools/perf/util/thread.h
index 1dfd9ff8bdcd..ee6bbcf277ca 100644
--- a/tools/perf/util/thread.h
+++ b/tools/perf/util/thread.h
@@ -6,7 +6,10 @@
 #include "symbol.h"
 
 struct thread {
-	struct rb_node		rb_node;
+	union {
+		struct rb_node	 rb_node;
+		struct list_head node;
+	};
 	struct map_groups	mg;
 	pid_t			pid;
 	char			shortname[3];