diff --git a/tensorflow/compiler/jit/BUILD b/tensorflow/compiler/jit/BUILD
index 682c0f0cb05..bfc0b0070ee 100644
--- a/tensorflow/compiler/jit/BUILD
+++ b/tensorflow/compiler/jit/BUILD
@@ -45,15 +45,37 @@ cc_library(
     alwayslink = 1,
 )
 
+cc_library(
+    name = "register_xla_cpu_jit",
+    srcs = ["register_xla_cpu_jit.cc"],
+    visibility = ["//visibility:public"],
+    deps = [
+        "//tensorflow/core:tf_xla_stub",
+    ],
+    alwayslink = 1,
+)
+
 cc_library(
     name = "xla_cpu_jit",
     visibility = ["//visibility:public"],
     deps = [
         ":jit_compilation_passes",
+        ":register_xla_cpu_jit",
         "//tensorflow/compiler/jit/kernels:xla_ops",
         "//tensorflow/compiler/tf2xla/kernels:xla_dummy_ops",
         "//tensorflow/compiler/tf2xla/kernels:xla_ops",
         "//tensorflow/compiler/xla/service:cpu_plugin",
+        "//tensorflow/core:tf_xla_stub",
+    ],
+    alwayslink = 1,
+)
+
+cc_library(
+    name = "register_xla_gpu_jit",
+    srcs = ["register_xla_gpu_jit.cc"],
+    visibility = ["//visibility:public"],
+    deps = [
+        "//tensorflow/core:tf_xla_stub",
     ],
     alwayslink = 1,
 )
@@ -63,6 +85,8 @@ cc_library(
     visibility = ["//visibility:public"],
     deps = if_cuda([
         ":jit_compilation_passes",
+        ":register_xla_gpu_jit",
+        "//tensorflow/core:tf_xla_stub",
         "//tensorflow/compiler/jit/kernels:xla_ops",
         "//tensorflow/compiler/tf2xla/kernels:xla_ops",
         "//tensorflow/compiler/tf2xla/kernels:xla_dummy_ops",
@@ -78,6 +102,7 @@ cc_library(
     deps = [
         ":flags",
         ":jit_compilation_passes",
+        ":register_xla_cpu_jit",
         ":xla_device",
         "//tensorflow/compiler/jit/kernels:xla_ops",
         "//tensorflow/compiler/tf2xla:xla_compiler",
@@ -96,6 +121,7 @@ cc_library(
     visibility = [":friends"],
     deps = [
         ":jit_compilation_passes",
+        ":register_xla_gpu_jit",
         ":xla_device",
         "//tensorflow/compiler/jit/kernels:xla_ops",
         "//tensorflow/compiler/tf2xla:xla_compiler",
diff --git a/tensorflow/compiler/jit/register_xla_cpu_jit.cc b/tensorflow/compiler/jit/register_xla_cpu_jit.cc
new file mode 100644
index 00000000000..9cbef631271
--- /dev/null
+++ b/tensorflow/compiler/jit/register_xla_cpu_jit.cc
@@ -0,0 +1,22 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+
+#include "tensorflow/core/common_runtime/tf_xla_stub.h"
+
+namespace tensorflow {
+namespace {
+XlaCpuJitIsLinkedIn register_xla_cpu_jit;
+}
+}  // namespace tensorflow
diff --git a/tensorflow/compiler/jit/register_xla_gpu_jit.cc b/tensorflow/compiler/jit/register_xla_gpu_jit.cc
new file mode 100644
index 00000000000..7399a41d25f
--- /dev/null
+++ b/tensorflow/compiler/jit/register_xla_gpu_jit.cc
@@ -0,0 +1,22 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+
+#include "tensorflow/core/common_runtime/tf_xla_stub.h"
+
+namespace tensorflow {
+namespace {
+XlaGpuJitIsLinkedIn register_xla_gpu_jit;
+}
+}  // namespace tensorflow
diff --git a/tensorflow/core/BUILD b/tensorflow/core/BUILD
index 73e8db58a83..1162c47e6bf 100644
--- a/tensorflow/core/BUILD
+++ b/tensorflow/core/BUILD
@@ -2976,6 +2976,19 @@ tf_cuda_library(
     ] + if_static([":core_cpu_impl"]) + tf_protos_all() + tf_protos_grappler(),
 )
 
+tf_cuda_library(
+    name = "tf_xla_stub",
+    srcs = ["common_runtime/tf_xla_stub.cc"],
+    hdrs = ["common_runtime/tf_xla_stub.h"],
+    copts = tf_copts(),
+    visibility = ["//visibility:public"],
+    deps = [
+        ":lib",
+        ":proto_text",
+        ":session_options",
+    ],
+)
+
 tf_cuda_library(
     name = "core_cpu_internal",
     srcs = [
@@ -2989,6 +3002,7 @@ tf_cuda_library(
         ":framework",
         ":graph",
         ":lib",
+        ":tf_xla_stub",
         ":proto_text",
         ":protos_all_cc",
         "//tensorflow/core/grappler:grappler_item",
diff --git a/tensorflow/core/common_runtime/graph_execution_state.cc b/tensorflow/core/common_runtime/graph_execution_state.cc
index 0d36930324a..c18268ad7bc 100644
--- a/tensorflow/core/common_runtime/graph_execution_state.cc
+++ b/tensorflow/core/common_runtime/graph_execution_state.cc
@@ -25,6 +25,7 @@ limitations under the License.
 #include "tensorflow/core/common_runtime/device.h"
 #include "tensorflow/core/common_runtime/optimization_registry.h"
 #include "tensorflow/core/common_runtime/placer.h"
+#include "tensorflow/core/common_runtime/tf_xla_stub.h"
 #include "tensorflow/core/framework/graph.pb_text.h"
 #include "tensorflow/core/framework/graph_def_util.h"
 #include "tensorflow/core/framework/node_def.pb.h"
@@ -720,6 +721,8 @@ Status GraphExecutionState::BuildGraph(const BuildGraphOptions& options,
   CHECK_EQ(options.callable_options.fetch_size(),
            rewrite_metadata.fetch_types.size());
 
+  TF_RETURN_IF_ERROR(CheckXlaJitOptimizerOptions(session_options_));
+
   // TODO(andydavis): Clarify optimization pass requirements around CostModel.
   GraphOptimizationPassOptions optimization_options;
   optimization_options.session_options = session_options_;
diff --git a/tensorflow/core/common_runtime/tf_xla_stub.cc b/tensorflow/core/common_runtime/tf_xla_stub.cc
new file mode 100644
index 00000000000..d463693669f
--- /dev/null
+++ b/tensorflow/core/common_runtime/tf_xla_stub.cc
@@ -0,0 +1,82 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+
+#include <cstdlib>
+
+#include "tensorflow/core/common_runtime/tf_xla_stub.h"
+#include "tensorflow/core/lib/core/errors.h"
+
+namespace tensorflow {
+namespace {
+
+bool is_xla_gpu_jit_registered = false;
+bool is_xla_cpu_jit_registered = false;
+
+struct XlaEnvVars {
+  bool xla_flags_env_var_present;
+  bool tf_xla_flags_env_var_present;
+};
+
+XlaEnvVars ComputeEnvVarHasXlaFlags() {
+  XlaEnvVars env_vars;
+  env_vars.xla_flags_env_var_present = getenv("XLA_FLAGS") != nullptr;
+  env_vars.tf_xla_flags_env_var_present = getenv("TF_XLA_FLAGS") != nullptr;
+  return env_vars;
+}
+
+}  // namespace
+
+XlaGpuJitIsLinkedIn::XlaGpuJitIsLinkedIn() { is_xla_gpu_jit_registered = true; }
+XlaCpuJitIsLinkedIn::XlaCpuJitIsLinkedIn() { is_xla_cpu_jit_registered = true; }
+
+Status CheckXlaJitOptimizerOptions(const SessionOptions* session_options) {
+  static XlaEnvVars env_vars = ComputeEnvVarHasXlaFlags();
+
+  if (is_xla_cpu_jit_registered || is_xla_gpu_jit_registered) {
+    return Status::OK();
+  }
+
+  if (env_vars.xla_flags_env_var_present) {
+    return errors::InvalidArgument(
+        "The XLA JIT is not linked in but the \"XLA_FLAGS\" environment "
+        "variable is set.  Please either link in XLA or remove \"XLA_FLAGS\" "
+        "from the environment.");
+  }
+
+  if (env_vars.tf_xla_flags_env_var_present) {
+    return errors::InvalidArgument(
+        "The XLA JIT is not linked in but the \"TF_XLA_FLAGS\" environment "
+        "variable is set.  Please either link in XLA or remove "
+        "\"TF_XLA_FLAGS\" from the environment.");
+  }
+
+  if (session_options) {
+    OptimizerOptions::GlobalJitLevel jit_level =
+        session_options->config.graph_options()
+            .optimizer_options()
+            .global_jit_level();
+
+    if (jit_level == OptimizerOptions::ON_1 ||
+        jit_level == OptimizerOptions::ON_2) {
+      return errors::InvalidArgument(
+          "The XLA JIT is enabled in the session options but XLA is not linked "
+          "in.  Plesae either link in XLA or disable the JIT in the session "
+          "options.");
+    }
+  }
+
+  return Status::OK();
+}
+}  // namespace tensorflow
diff --git a/tensorflow/core/common_runtime/tf_xla_stub.h b/tensorflow/core/common_runtime/tf_xla_stub.h
new file mode 100644
index 00000000000..723b2b5cd2e
--- /dev/null
+++ b/tensorflow/core/common_runtime/tf_xla_stub.h
@@ -0,0 +1,50 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+#ifndef TENSORFLOW_CORE_COMMON_RUNTIME_TF_XLA_STUB_H_
+#define TENSORFLOW_CORE_COMMON_RUNTIME_TF_XLA_STUB_H_
+
+#include "tensorflow/core/lib/core/status.h"
+#include "tensorflow/core/public/session_options.h"
+
+namespace tensorflow {
+// Returns an error if the XLA JIT is enabled via `session_options` or if the
+// TF_XLA_FLAGS or XLA_FLAGS environment variables are set, but neither the
+// XLA CPU JIT nor the XLA GPU JIT are linked in.
+//
+// If `session_options` is null then only the environment variables are checked.
+Status CheckXlaJitOptimizerOptions(const SessionOptions* session_options);
+
+// The XLA CPU JIT creates a static instance of this class to notify
+// `CheckXlaJitOptimizerOptions` that the XLA CPU JIT is linked in.
+//
+// NB!  The constructor of this class (if run at all) needs to be ordered (via
+// happens before) before any call to `CheckXlaJitOptimizerOptions`.
+class XlaCpuJitIsLinkedIn {
+ public:
+  XlaCpuJitIsLinkedIn();
+};
+
+// The XLA GPU JIT creates a static instance of this class to notify
+// `CheckXlaJitOptimizerOptions` that the XLA GPU JIT is linked in.
+//
+// NB!  The constructor of this class (if run at all) needs to be ordered (via
+// happens before) before any call to `CheckXlaJitOptimizerOptions`.
+class XlaGpuJitIsLinkedIn {
+ public:
+  XlaGpuJitIsLinkedIn();
+};
+}  // namespace tensorflow
+
+#endif  // TENSORFLOW_CORE_COMMON_RUNTIME_TF_XLA_STUB_H_