diff --git a/sycl/test-e2e/Basic/queue/release.cpp b/sycl/test-e2e/Basic/queue/release.cpp
index 1cf717a63c764..aef5160123ce4 100644
--- a/sycl/test-e2e/Basic/queue/release.cpp
+++ b/sycl/test-e2e/Basic/queue/release.cpp
@@ -1,8 +1,7 @@
 // RUN: %{build} -o %t.out
-// RUN: env SYCL_UR_TRACE=2 %{run} %t.out | FileCheck %s
+// RUN: env SYCL_UR_TRACE=2 %{run} %t.out | FileCheck %s %if !windows %{--check-prefixes=CHECK-RELEASE%}
 //
-// TODO: Reenable on Windows, see https://github.com/intel/llvm/issues/14768
-// XFAIL: hip_nvidia, windows
+// XFAIL: hip_nvidia
 
 #include <sycl/detail/core.hpp>
 int main() {
@@ -19,7 +18,14 @@ int main() {
 // specific queue workaround.
 // CHECK-DAG: <--- urEventRelease(
 // CHECK-DAG: <--- urQueueRelease(
-// CHECK: <--- urContextRelease(
-// CHECK: <--- urKernelRelease(
-// CHECK: <--- urProgramRelease(
-// CHECK: <--- urDeviceRelease(
+
+// On Windows, dlls unloading is inconsistent and if we try to release these UR
+// objects manually, inconsistent hangs happen due to a race between unloading
+// the UR adapters dlls (in addition to their dependency dlls) and the releasing
+// of these UR objects. So, we currently shutdown without releasing them and
+// windows should handle the memory cleanup.
+
+// CHECK-RELEASE: <--- urContextRelease(
+// CHECK-RELEASE: <--- urKernelRelease(
+// CHECK-RELEASE: <--- urProgramRelease(
+// CHECK-RELEASE: <--- urDeviceRelease(
diff --git a/sycl/test-e2e/KernelAndProgram/disable-caching.cpp b/sycl/test-e2e/KernelAndProgram/disable-caching.cpp
index 01031abd84be9..ae3b5e435dd4a 100644
--- a/sycl/test-e2e/KernelAndProgram/disable-caching.cpp
+++ b/sycl/test-e2e/KernelAndProgram/disable-caching.cpp
@@ -5,10 +5,7 @@
 // RUN: env ZE_DEBUG=-6 SYCL_UR_TRACE=2 SYCL_CACHE_IN_MEM=0 %{run} %t.out \
 // RUN: | FileCheck %s
 // RUN: env ZE_DEBUG=-6 SYCL_UR_TRACE=2 %{run} %t.out \
-// RUN: | FileCheck %s --check-prefixes=CHECK-CACHE
-
-// TODO: Reenable on Windows, see https://github.com/intel/llvm/issues/14768
-// XFAIL: windows
+// RUN: | FileCheck %s --check-prefixes=CHECK-CACHE%if !windows %{,CHECK-RELEASE%}
 
 #include <sycl/detail/core.hpp>
 
@@ -93,10 +90,16 @@ int main() {
   free(p, q);
 }
 
+// On Windows, dlls unloading is inconsistent and if we try to release these UR
+// objects manually, inconsistent hangs happen due to a race between unloading
+// the UR adapters dlls (in addition to their dependency dlls) and the releasing
+// of these UR objects. So, we currently shutdown without releasing them and
+// windows should handle the memory cleanup.
+
 // (Program cache releases)
-// CHECK-CACHE: <--- urKernelRelease
-// CHECK-CACHE: <--- urKernelRelease
-// CHECK-CACHE: <--- urKernelRelease
-// CHECK-CACHE: <--- urProgramRelease
-// CHECK-CACHE: <--- urProgramRelease
-// CHECK-CACHE: <--- urProgramRelease
+// CHECK-RELEASE: <--- urKernelRelease
+// CHECK-RELEASE: <--- urKernelRelease
+// CHECK-RELEASE: <--- urKernelRelease
+// CHECK-RELEASE: <--- urProgramRelease
+// CHECK-RELEASE: <--- urProgramRelease
+// CHECK-RELEASE: <--- urProgramRelease
diff --git a/sycl/test-e2e/Regression/context_is_destroyed_after_exception.cpp b/sycl/test-e2e/Regression/context_is_destroyed_after_exception.cpp
index 2d88325b86293..57e48df0c664a 100644
--- a/sycl/test-e2e/Regression/context_is_destroyed_after_exception.cpp
+++ b/sycl/test-e2e/Regression/context_is_destroyed_after_exception.cpp
@@ -1,10 +1,9 @@
 // REQUIRES: gpu
 
 // RUN: %{build} -o %t.out
-// RUN: env SYCL_UR_TRACE=2 %{run} %t.out 2>&1 | FileCheck %s
+// RUN: env SYCL_UR_TRACE=2 %{run} %t.out %if !windows %{2>&1 | FileCheck %s %}
 //
-// TODO: Reenable on Windows, see https://github.com/intel/llvm/issues/14768
-// XFAIL: hip_nvidia, windows
+// XFAIL: hip_nvidia
 
 #include <sycl/detail/core.hpp>
 
@@ -32,4 +31,10 @@ int main() {
   return 0;
 }
 
+// On Windows, dlls unloading is inconsistent and if we try to release these UR
+// objects manually, inconsistent hangs happen due to a race between unloading
+// the UR adapters dlls (in addition to their dependency dlls) and the releasing
+// of these UR objects. So, we currently shutdown without releasing them and
+// windows should handle the memory cleanup.
+
 // CHECK: <--- urContextRelease(
diff --git a/sycl/test-e2e/Regression/pi_release.cpp b/sycl/test-e2e/Regression/pi_release.cpp
index 3a8587f99af0b..81ca25ddf6233 100644
--- a/sycl/test-e2e/Regression/pi_release.cpp
+++ b/sycl/test-e2e/Regression/pi_release.cpp
@@ -1,9 +1,6 @@
 // REQUIRES: opencl || level_zero || cuda
 // RUN: %{build} -o %t.out
-// RUN: env SYCL_UR_TRACE=2 %{run} %t.out 2>&1 | FileCheck %s
-//
-// TODO: Reenable on Windows, see https://github.com/intel/llvm/issues/14768
-// XFAIL: windows
+// RUN: env SYCL_UR_TRACE=2 %{run} %t.out %if !windows %{2>&1 | FileCheck %s %}
 
 #include <sycl/detail/core.hpp>
 
@@ -12,5 +9,11 @@ int main() {
   return 0;
 }
 
+// On Windows, dlls unloading is inconsistent and if we try to release these UR
+// objects manually, inconsistent hangs happen due to a race between unloading
+// the UR adapters dlls (in addition to their dependency dlls) and the releasing
+// of these UR objects. So, we currently shutdown without releasing them and
+// windows should handle the memory cleanup.
+
 // CHECK: <--- urQueueRelease
 // CHECK: <--- urContextRelease
diff --git a/sycl/test-e2e/Scheduler/ReleaseResourcesTest.cpp b/sycl/test-e2e/Scheduler/ReleaseResourcesTest.cpp
index 589df5fdcce82..880ec728f8951 100644
--- a/sycl/test-e2e/Scheduler/ReleaseResourcesTest.cpp
+++ b/sycl/test-e2e/Scheduler/ReleaseResourcesTest.cpp
@@ -1,8 +1,7 @@
 // RUN: %{build} -Wno-error=unused-command-line-argument -fsycl-dead-args-optimization -o %t.out
-// RUN: env SYCL_UR_TRACE=2 %{run} %t.out 2>&1 | FileCheck %s
+// RUN: env SYCL_UR_TRACE=2 %{run} %t.out 2>&1 | FileCheck %s %if !windows %{--check-prefix=CHECK-RELEASE%}
 //
-// TODO: Reenable on Windows, see https://github.com/intel/llvm/issues/14768
-// XFAIL: hip_nvidia, windows
+// XFAIL: hip_nvidia
 
 //==------------------- ReleaseResourcesTests.cpp --------------------------==//
 //
@@ -50,7 +49,14 @@ int main() {
 // CHECK: <--- urQueueCreate
 // CHECK: <--- urProgramCreate
 // CHECK: <--- urKernelCreate
-// CHECK: <--- urQueueRelease
-// CHECK: <--- urContextRelease
-// CHECK: <--- urKernelRelease
-// CHECK: <--- urProgramRelease
+
+// On Windows, dlls unloading is inconsistent and if we try to release these UR
+// objects manually, inconsistent hangs happen due to a race between unloading
+// the UR adapters dlls (in addition to their dependency dlls) and the releasing
+// of these UR objects. So, we currently shutdown without releasing them and
+// windows should handle the memory cleanup.
+
+// CHECK-RELEASE: <--- urQueueRelease
+// CHECK-RELEASE: <--- urContextRelease
+// CHECK-RELEASE: <--- urKernelRelease
+// CHECK-RELEASE: <--- urProgramRelease