diff --git a/src/d3d11/d3d11_swapchain.cpp b/src/d3d11/d3d11_swapchain.cpp
index cb5679854..e2100eb8d 100644
--- a/src/d3d11/d3d11_swapchain.cpp
+++ b/src/d3d11/d3d11_swapchain.cpp
@@ -295,6 +295,9 @@ namespace dxvk {
     if (m_latencyHud)
       m_latencyHud->accumulateStats(latencyStats);
 
+    if (m_renderLatencyHud)
+      m_renderLatencyHud->updateLatencyTracker(m_latency);
+
     return hr;
   }
 
@@ -604,8 +607,14 @@ namespace dxvk {
     if (hud) {
       hud->addItem<hud::HudClientApiItem>("api", 1, GetApiName());
 
-      if (m_latency)
+      if (m_latency) {
         m_latencyHud = hud->addItem<hud::HudLatencyItem>("latency", 4);
+        FramePacer* framePacer = dynamic_cast<FramePacer*>(m_latency.ptr());
+        if (framePacer) {
+          int32_t fpsItemPos = hud->getItemPos<hud::HudFpsItem>();
+          m_renderLatencyHud = hud->addItem<hud::HudRenderLatencyItem>("renderlatency", fpsItemPos+1);
+        }
+      }
     }
 
     m_blitter = new DxvkSwapchainBlitter(m_device, std::move(hud));
diff --git a/src/d3d11/d3d11_swapchain.h b/src/d3d11/d3d11_swapchain.h
index 99f09450c..6a77c7351 100644
--- a/src/d3d11/d3d11_swapchain.h
+++ b/src/d3d11/d3d11_swapchain.h
@@ -125,7 +125,8 @@ namespace dxvk {
     dxvk::mutex               m_frameStatisticsLock;
     DXGI_VK_FRAME_STATISTICS  m_frameStatistics = { };
 
-    Rc<hud::HudLatencyItem>   m_latencyHud;
+    Rc<hud::HudLatencyItem>       m_latencyHud;
+    Rc<hud::HudRenderLatencyItem> m_renderLatencyHud;
 
     Rc<DxvkImageView> GetBackBufferView();
 
diff --git a/src/d3d9/d3d9_swapchain.cpp b/src/d3d9/d3d9_swapchain.cpp
index 05466eca5..539b0815c 100644
--- a/src/d3d9/d3d9_swapchain.cpp
+++ b/src/d3d9/d3d9_swapchain.cpp
@@ -925,6 +925,9 @@ namespace dxvk {
     if (m_latencyHud)
       m_latencyHud->accumulateStats(latencyStats);
 
+    if (m_renderLatencyHud)
+      m_renderLatencyHud->updateLatencyTracker(m_latencyTracker);
+
     // Rotate swap chain buffers so that the back
     // buffer at index 0 becomes the front buffer.
     for (uint32_t i = 1; i < m_backBuffers.size(); i++)
@@ -1062,8 +1065,14 @@ namespace dxvk {
     if (hud) {
       m_apiHud = hud->addItem<hud::HudClientApiItem>("api", 1, GetApiName());
 
-      if (m_latencyTracking)
+      if (m_latencyTracking) {
         m_latencyHud = hud->addItem<hud::HudLatencyItem>("latency", 4);
+        FramePacer* framePacer = dynamic_cast<FramePacer*>(m_latencyTracker.ptr());
+        if (framePacer) {
+          int32_t fpsItemPos = hud->getItemPos<hud::HudFpsItem>();
+          m_renderLatencyHud = hud->addItem<hud::HudRenderLatencyItem>("renderlatency", fpsItemPos+1);
+        }
+      }
 
       hud->addItem<hud::HudSamplerCount>("samplers", -1, m_parent);
       hud->addItem<hud::HudFixedFunctionShaders>("ffshaders", -1, m_parent);
diff --git a/src/d3d9/d3d9_swapchain.h b/src/d3d9/d3d9_swapchain.h
index 6ea0d96cb..d06c388a9 100644
--- a/src/d3d9/d3d9_swapchain.h
+++ b/src/d3d9/d3d9_swapchain.h
@@ -183,8 +183,9 @@ namespace dxvk {
     bool                      m_latencyTracking = false;
     Rc<DxvkLatencyTracker>    m_latencyTracker = nullptr;
 
-    Rc<hud::HudClientApiItem> m_apiHud;
-    Rc<hud::HudLatencyItem>   m_latencyHud;
+    Rc<hud::HudClientApiItem>     m_apiHud;
+    Rc<hud::HudLatencyItem>       m_latencyHud;
+    Rc<hud::HudRenderLatencyItem> m_renderLatencyHud;
 
     std::optional<VkHdrMetadataEXT> m_hdrMetadata;
     bool m_unlockAdditionalFormats = false;
diff --git a/src/dxvk/hud/dxvk_hud.h b/src/dxvk/hud/dxvk_hud.h
index 58c383f07..388cbf22a 100644
--- a/src/dxvk/hud/dxvk_hud.h
+++ b/src/dxvk/hud/dxvk_hud.h
@@ -59,6 +59,11 @@ namespace dxvk::hud {
     Rc<T> addItem(const char* name, int32_t at, Args... args) {
       return m_hudItems.add<T>(name, at, std::forward<Args>(args)...);
     }
+
+    template<typename T>
+    int32_t getItemPos() {
+      return m_hudItems.getItemPos<T>();
+    }
     
     /**
      * \brief Creates the HUD
diff --git a/src/dxvk/hud/dxvk_hud_item.cpp b/src/dxvk/hud/dxvk_hud_item.cpp
index f80ba77d4..138f1333f 100644
--- a/src/dxvk/hud/dxvk_hud_item.cpp
+++ b/src/dxvk/hud/dxvk_hud_item.cpp
@@ -1,4 +1,5 @@
 #include "dxvk_hud_item.h"
+#include "../framepacer/dxvk_framepacer.h"
 
 #include <hud_chunk_frag_background.h>
 #include <hud_chunk_frag_visualize.h>
@@ -213,6 +214,63 @@ namespace dxvk::hud {
   }
 
 
+  HudRenderLatencyItem::HudRenderLatencyItem() { }
+  HudRenderLatencyItem::~HudRenderLatencyItem() { }
+
+  void HudRenderLatencyItem::update(dxvk::high_resolution_clock::time_point time) {
+    // we cannot measure latency when fps-limiting is performed in Presenter::runFrameThread()
+    // because it's interfering with getting the right timestamp from vkWaitForPresent()
+    // if we truely wanted to measure it, we would need one additional thread
+    if (FpsLimiter::m_isActive) {
+      m_latency = "N/A";
+      return;
+    }
+
+    const Rc<DxvkLatencyTracker> tracker = m_tracker;
+    const FramePacer* framePacer = dynamic_cast<FramePacer*>( tracker.ptr() );
+    if (!framePacer)
+      return;
+
+    auto elapsed = std::chrono::duration_cast<std::chrono::microseconds>(time - m_lastUpdate);
+
+    if (elapsed.count() >= UpdateInterval) {
+      m_lastUpdate = time;
+
+      LatencyMarkersReader reader = framePacer->m_latencyMarkersStorage.getReader(100);
+      const LatencyMarkers* markers;
+      uint32_t count = 0;
+      uint64_t totalLatency = 0;
+      while (reader.getNext(markers)) {
+        totalLatency += markers->presentFinished;
+        ++count;
+      }
+
+      if (!count)
+        return;
+
+      uint64_t latency = totalLatency / count;
+      m_latency = str::format(latency / 1000, ".", (latency/100) % 10, " ms");
+    }
+  }
+
+
+  HudPos HudRenderLatencyItem::render(
+    const DxvkContextObjects& ctx,
+    const HudPipelineKey&     key,
+    const HudOptions&         options,
+          HudRenderer&        renderer,
+          HudPos              position) {
+
+    position.y += 12;
+    renderer.drawText(16, position, 0xff4040ffu, "Render latency:");
+    renderer.drawText(16, { position.x + 195, position.y },
+      0xffffffffu, m_latency);
+
+    position.y += 8;
+    return position;
+  }
+
+
   HudFrameTimeItem::HudFrameTimeItem(const Rc<DxvkDevice>& device, HudRenderer* renderer)
   : m_device            (device),
     m_gfxSetLayout      (createDescriptorSetLayout()),
diff --git a/src/dxvk/hud/dxvk_hud_item.h b/src/dxvk/hud/dxvk_hud_item.h
index 866849b4e..fab3fc97f 100644
--- a/src/dxvk/hud/dxvk_hud_item.h
+++ b/src/dxvk/hud/dxvk_hud_item.h
@@ -131,6 +131,15 @@ namespace dxvk::hud {
       return value;
     }
 
+    template<typename T>
+    int32_t getItemPos() {
+      for (int i=0; i<(int)m_items.size(); ++i) {
+        if (dynamic_cast<T*>(m_items[i].ptr()))
+          return i;
+      }
+      return -1;
+    }
+
   private:
 
     bool                                          m_enableFull = false;
@@ -244,6 +253,42 @@ namespace dxvk::hud {
   };
 
 
+   /**
+   * \brief HUD item to display render latency
+   */
+  class HudRenderLatencyItem : public HudItem {
+    constexpr static int64_t UpdateInterval = 500'000;
+  public:
+
+    HudRenderLatencyItem();
+
+    ~HudRenderLatencyItem();
+
+    void updateLatencyTracker( const Rc<DxvkLatencyTracker>& tracker ) {
+      m_tracker = tracker;
+    }
+
+    void update(dxvk::high_resolution_clock::time_point time);
+
+    HudPos render(
+      const DxvkContextObjects& ctx,
+      const HudPipelineKey&     key,
+      const HudOptions&         options,
+            HudRenderer&        renderer,
+            HudPos              position);
+
+  private:
+
+    Rc<DxvkLatencyTracker> m_tracker;
+
+    dxvk::high_resolution_clock::time_point m_lastUpdate
+      = dxvk::high_resolution_clock::now();
+
+    std::string m_latency;
+
+  };
+
+
   /**
    * \brief HUD item to display the frame rate
    */