diff --git a/fluidsynth/src/rvoice/fluid_rvoice_event.c b/fluidsynth/src/rvoice/fluid_rvoice_event.c
index a62152fa..ffe28d97 100644
--- a/fluidsynth/src/rvoice/fluid_rvoice_event.c
+++ b/fluidsynth/src/rvoice/fluid_rvoice_event.c
@@ -233,7 +233,13 @@ new_fluid_rvoice_eventhandler(int is_threadsafe, int queuesize,
   eventhandler->mixer = NULL;
   eventhandler->queue = NULL;
   eventhandler->finished_voices = NULL;
-  eventhandler->is_threadsafe = is_threadsafe;
+  
+  /* HACK 2017-08-27: always enforce threadsafety, i.e. enforce enqueuing events
+   * otherwise we mess up rendering if more than one block is requested by the user
+   * because fluid_rvoice_eventhandler_dispatch_count() always stays zero causing
+   * that too many events are dispatched too early, causing incorrectly timed audio
+   */
+  eventhandler->is_threadsafe = TRUE;
   eventhandler->queue_stored = 0;
   
   eventhandler->finished_voices = new_fluid_ringbuffer(finished_voices_size,
diff --git a/fluidsynth/src/synth/fluid_synth.c b/fluidsynth/src/synth/fluid_synth.c
index 652252ec..98c9c510 100644
--- a/fluidsynth/src/synth/fluid_synth.c
+++ b/fluidsynth/src/synth/fluid_synth.c
@@ -2939,17 +2939,10 @@ fluid_synth_render_blocks(fluid_synth_t* synth, int blockcount)
   
   fluid_rvoice_eventhandler_dispatch_all(synth->eventhandler);
   
-#if 0
   /* do not render more blocks than we can store internally */
   maxblocks = fluid_rvoice_mixer_get_bufcount(synth->eventhandler->mixer);
   if (blockcount > maxblocks)
       blockcount = maxblocks;
-#else
-  /* if we process more blocks, noteons will be timed incorrectly
-   * this bug seems to be design made, i.e. the current implementation requires to
-   * process all events for one block and then render whatever we just processed. */
-  blockcount = 1;
-#endif
   
   for (i=0; i < blockcount; i++) {
     fluid_sample_timer_process(synth);