diff --git a/tgt-vvp/vvp_process.c b/tgt-vvp/vvp_process.c
index 44731fc61..e4138f8cc 100644
--- a/tgt-vvp/vvp_process.c
+++ b/tgt-vvp/vvp_process.c
@@ -1728,12 +1728,7 @@ static int show_stmt_wait(ivl_statement_t net, ivl_scope_t sscope)
       if ((ivl_stmt_nevent(net) == 1) && (ivl_stmt_events(net, 0) == 0)) {
 	    assert(ivl_statement_type(ivl_stmt_sub_stmt(net)) == IVL_ST_NOOP);
 	    show_stmt_file_line(net, "Wait fork statement.");
-	    fprintf(vvp_out, "    %%wait_fork;\n");
-
-	    fprintf(stderr, "%s:%u: vvp.tgt sorry: wait fork is not currently "
-			    "supported.\n",
-			    ivl_stmt_file(net), ivl_stmt_lineno(net));
-	    vvp_errors += 1;
+	    fprintf(vvp_out, "    %%wait/fork;\n");
 	    return 0;
       }
 
diff --git a/vvp/codes.h b/vvp/codes.h
index 709db5495..3cab77057 100644
--- a/vvp/codes.h
+++ b/vvp/codes.h
@@ -209,6 +209,7 @@ extern bool of_SUBSTR_V(vthread_t thr, vvp_code_t code);
 extern bool of_TEST_NUL(vthread_t thr, vvp_code_t code);
 extern bool of_VPI_CALL(vthread_t thr, vvp_code_t code);
 extern bool of_WAIT(vthread_t thr, vvp_code_t code);
+extern bool of_WAIT_FORK(vthread_t thr, vvp_code_t code);
 extern bool of_XNOR(vthread_t thr, vvp_code_t code);
 extern bool of_XNORR(vthread_t thr, vvp_code_t code);
 extern bool of_XOR(vthread_t thr, vvp_code_t code);
diff --git a/vvp/compile.cc b/vvp/compile.cc
index c5c6857d0..a19264495 100644
--- a/vvp/compile.cc
+++ b/vvp/compile.cc
@@ -255,6 +255,7 @@ static const struct opcode_table_s opcode_table[] = {
       { "%substr/v",of_SUBSTR_V,3,{OA_BIT1,     OA_BIT2,     OA_NUMBER} },
       { "%test_nul",  of_TEST_NUL, 1,{OA_FUNC_PTR,OA_NONE,     OA_NONE} },
       { "%wait",   of_WAIT,   1,  {OA_FUNC_PTR, OA_NONE,     OA_NONE} },
+      { "%wait/fork",of_WAIT_FORK,0,{OA_NONE,   OA_NONE,     OA_NONE} },
       { "%xnor",   of_XNOR,   3,  {OA_BIT1,     OA_BIT2,     OA_NUMBER} },
       { "%xnor/r", of_XNORR,  3,  {OA_BIT1,     OA_BIT2,     OA_NUMBER} },
       { "%xor",    of_XOR,    3,  {OA_BIT1,     OA_BIT2,     OA_NUMBER} },
diff --git a/vvp/opcodes.txt b/vvp/opcodes.txt
index e6c5bc08f..b7bcc2dc2 100644
--- a/vvp/opcodes.txt
+++ b/vvp/opcodes.txt
@@ -1159,6 +1159,12 @@ threads that await the functor. When the defined sort of event occurs
 on the functor, a thread schedule event is created for all the threads
 in its list and the list is cleared.
 
+* %wait/fork
+
+This instruction puts the current thread to sleep until all the detached
+children have finished executing. The last detached child is responsible
+for restarting the parent when it finishes.
+
 * %xnor <dst>, <src>, <wid>
 
 This does a bitwise exclusive nor (~^) of the <src> and <dst> vector,
diff --git a/vvp/vthread.cc b/vvp/vthread.cc
index 4895b5a44..eb02532f8 100644
--- a/vvp/vthread.cc
+++ b/vvp/vthread.cc
@@ -204,6 +204,7 @@ struct vthread_s {
 	/* My parent sets this when it wants me to wake it up. */
       unsigned i_am_joining      :1;
       unsigned i_am_detached     :1;
+      unsigned i_am_waiting      :1;
       unsigned i_have_ended      :1;
       unsigned waiting_for_event :1;
       unsigned is_scheduled      :1;
@@ -543,6 +544,7 @@ vthread_t vthread_new(vvp_code_t pc, struct __vpiScope*scope)
 
       thr->i_am_joining  = 0;
       thr->i_am_detached = 0;
+      thr->i_am_waiting  = 0;
       thr->is_scheduled  = 0;
       thr->i_have_ended  = 0;
       thr->delay_delete  = 0;
@@ -2604,11 +2606,22 @@ bool of_END(vthread_t thr, vvp_code_t)
       }
 
 	/* If this thread is not fully detached then remove it from the
-	 * parents detached_children set. */
+	 * parents detached_children set and reap it. */
       if (thr->i_am_detached) {
-	    assert(thr->parent);
-	    size_t res = thr->parent->detached_children.erase(thr);
+	    vthread_t tmp = thr->parent;
+	    assert(tmp);
+	    size_t res = tmp->detached_children.erase(thr);
 	    assert(res == 1);
+	      /* If the parent is waiting for the detached children to
+	       * finish then the last detached child needs to tell the
+	       * parent to wake up when it is finished. */
+	    if (tmp->i_am_waiting && tmp->detached_children.empty()) {
+		  tmp->i_am_waiting = 0;
+		  schedule_vthread(tmp, 0, true);
+	    }
+	      /* Fully detach this thread so it will be reaped below. */
+	    thr->i_am_detached = 0;
+	    thr->parent = 0;
       }
 
 	/* If I have no parent, then no one can %join me and there is
@@ -5590,6 +5603,30 @@ bool of_WAIT(vthread_t thr, vvp_code_t cp)
       return false;
 }
 
+/*
+ * Implement the %wait/fork (SystemVerilog) instruction by suspending
+ * the current thread until all the detached children have finished.
+ */
+bool of_WAIT_FORK(vthread_t thr, vvp_code_t)
+{
+	/* If a %wait/fork is being executed then the parent thread
+	 * cannot be waiting in a join or already waiting. */
+      assert(! thr->i_am_joining);
+      assert(! thr->i_am_waiting);
+
+	/* There should be no active children when waiting. */
+      assert(thr->children.empty());
+
+	/* If there are no detached children then there is nothing to
+	 * wait for. */
+      if (thr->detached_children.empty()) return true;
+
+	/* Flag that this process is waiting for the detached children
+	 * to finish and suspend it. */
+      thr->i_am_waiting = 1;
+      return false;
+}
+
 
 bool of_XNOR(vthread_t thr, vvp_code_t cp)
 {