diff --git a/recipes/onnxruntime/all/CMakeLists.txt b/recipes/onnxruntime/all/CMakeLists.txt
new file mode 100644
index 00000000000000..1848ca5a77c355
--- /dev/null
+++ b/recipes/onnxruntime/all/CMakeLists.txt
@@ -0,0 +1,7 @@
+cmake_minimum_required(VERSION 2.8.12)
+project(cmake_wrapper)
+
+include(conanbuildinfo.cmake)
+conan_basic_setup()
+
+add_subdirectory("source_subfolder")
diff --git a/recipes/onnxruntime/all/conandata.yml b/recipes/onnxruntime/all/conandata.yml
new file mode 100644
index 00000000000000..28ba0fd5e36dcd
--- /dev/null
+++ b/recipes/onnxruntime/all/conandata.yml
@@ -0,0 +1,14 @@
+sources:
+  "1.7.1":
+    sha256: e24ff9a2a21dab74e6dfa821e9a267baa10e9423ffae5f5e404e3a2949eb41e2
+    url: https://github.com/microsoft/onnxruntime/archive/v1.7.1.tar.gz
+  "1.5.3":
+    sha256: 40fc79f8a3126caba5fcdced260439c7b18c72214fb8f7268d87fac22098dfba
+    url: https://github.com/microsoft/onnxruntime/archive/v1.5.3.tar.gz
+  "1.2.0":
+    sha256: 1ed2a4303d621682c42ebaf99fb9a8ecdc386035b15925b148aa56944abc38c8
+    url: https://github.com/microsoft/onnxruntime/archive/v1.2.0.tar.gz
+patches:
+  "1.7.1": []
+  "1.5.3": []
+  "1.2.0": []
diff --git a/recipes/onnxruntime/all/conanfile.py b/recipes/onnxruntime/all/conanfile.py
new file mode 100644
index 00000000000000..bbe3cbbc5069cd
--- /dev/null
+++ b/recipes/onnxruntime/all/conanfile.py
@@ -0,0 +1,227 @@
+import os
+from conans import ConanFile, tools
+
+
+class OnnxRuntimeConan(ConanFile):
+    name = "onnxruntime"
+    description = "ONNX Runtime: cross-platform, high performance scoring " \
+                  "engine for ML models https://aka.ms/onnxruntime"
+    license = "MIT License"
+    topics = (
+        "deep-learning",
+        "onnx",
+        "neural-networks",
+        "machine-learning",
+        "ai-framework",
+        "hardware-acceleration"
+    )
+    homepage = "https://www.onnxruntime.ai/"
+    url = "https://github.com/conan-io/conan-center-index"
+
+    settings = "os", "compiler", "build_type", "arch", "os_build"
+
+    options = {
+        "shared": [True, False],
+        "parallel": [True, False],
+        "force_min_size_rel": [True, False],
+        "with_dml": [True, False],
+        "with_openmp": [True, False],
+        "with_dnnl": [True, False],
+        "with_nnapi": [True, False],
+        "with_winml": [True, False],
+        "with_python_wheel": [True, False],
+        "with_csharp": [True, False],
+        "with_java": [True, False],
+        "with_tests": [True, False]
+    }
+
+    default_options = {k: False for k, v in options.items()}
+
+    @property
+    def _source_subfolder(self):
+        return "source_subfolder"
+
+    def build_requirements(self):
+        self.build_requires("protobuf/3.11.3")
+
+    def source(self):
+        # Unfortunatelly git do not pack submodules, so code below doesn't work
+        #
+        # tools.get(**self.conan_data["sources"][self.version])
+        # os.rename(f"onnxruntime-{self.version}", self._source_subfolder)
+        #
+        # Using tools.Git to get it:
+        git = tools.Git(folder=self._source_subfolder)
+        git.clone("https://github.com/microsoft/onnxruntime.git")
+        git.checkout(f"v{self.version}", submodule="recursive")
+
+    def build(self):
+        is_windows = self.settings.os_build == "Windows"
+        build_script = ".\\build.bat" if is_windows else "./build.sh"
+        build_args = ["--skip_submodule_sync"]
+
+        if self.options.force_min_size_rel:
+            build_args.extend(["--config", "MinSizeRel"])
+        elif self.settings.build_type is not None:
+            build_args.extend(["--config", str(self.settings.build_type)])
+
+        if self.options.shared:
+            build_args.append("--build_shared_lib")
+
+        if self.options.parallel:
+            build_args.append("--parallel")
+
+        if self.options.with_openmp:
+            build_args.append("--use_openmp")
+
+        if self.options.with_dnnl:
+            build_args.append("--use_dnnl")
+
+        if self.options.with_nnapi:
+            build_args.append("--use_nnapi")
+
+        if self.options.with_winml:
+            build_args.append("--use_winml")
+
+        if self.options.with_dml:
+            build_args.append("--use_dml")
+
+        if self.options.with_python_wheel:
+            build_args.append("--build_wheel")
+
+        if self.options.with_csharp:
+            build_args.append("--build_csharp")
+
+        if self.options.with_java:
+            build_args.append("--use_java")
+
+        if self.options.with_tests:
+            build_args.append("--tests")
+        else:
+            build_args.append("--skip_tests")
+
+        build_args.extend([
+            "--cmake_extra_defines",
+            f"CMAKE_INSTALL_PREFIX={self.package_folder}"
+        ])
+
+        # TODO full list of options in
+        # https://github.com/microsoft/onnxruntime/blob/master/BUILD.md
+
+        if self.settings.os == "Android":
+            build_args.append("--android")
+
+            android_ndk_root = None
+            android_sdk_root = None
+            if "ANDROID_NDK_ROOT" in os.environ:
+                android_ndk_root = os.environ.get("ANDROID_NDK_ROOT")
+            elif "ANDROID_NDK_HOME" in os.environ:
+                android_ndk_root = os.environ.get("ANDROID_NDK_HOME")
+
+            if "ANDROID_SDK_ROOT" in os.environ:
+                android_sdk_root = os.environ.get("ANDROID_SDK_ROOT")
+            elif "ANDROID_SDK_HOME" in os.environ:
+                android_sdk_root = os.environ.get("ANDROID_SDK_HOME")
+            elif "ANDROID_HOME" in os.environ:
+                android_sdk_root = os.environ.get("ANDROID_HOME")
+
+            if android_ndk_root is None:
+                ndk_bundle_path = os.path.join(android_sdk_root, "ndk-bundle")
+                if os.path.isdir(ndk_bundle_path):
+                    android_ndk_root = ndk_bundle_path
+
+            if android_ndk_root is None:
+                raise Exception("ANDROID_NDK_ROOT env not defined")
+            if android_sdk_root is None:
+                raise Exception("ANDROID_SDK_ROOT env not defined")
+
+            build_args.extend(["--android_sdk_path", android_sdk_root])
+            build_args.extend(["--android_ndk_path", android_ndk_root])
+            build_args.extend([
+                "--android_abi",
+                tools.to_android_abi(self.settings.arch)
+            ])
+
+            if self.settings.os.api_level:
+                build_args.extend([
+                    "--android_api", str(self.settings.os.api_level)
+                ])
+
+        with tools.chdir(self._source_subfolder):
+            self.run(" ".join([build_script] + build_args))
+
+    def package(self):
+        build_dir = self._onnxruntime_build_dir
+        self.run(f"cmake --build {build_dir} --target install")
+
+        if not self.options.shared:
+            self.copy(pattern="*.a", dst="lib", src=build_dir, keep_path=False)
+
+        self.copy(
+            "onnxruntime_perf_test",
+            dst="bin",
+            src=build_dir,
+            keep_path=False
+        )
+
+        providers_inc = "include/onnxruntime/core/providers"
+        for provider in ["nnapi", "dnnl"]:
+            if getattr(self.options, f"with_{provider}"):
+                dst = os.path.join(providers_inc, provider)
+                self.copy(
+                    pattern="*.h",
+                    dst=dst,
+                    src=os.path.join(self.name, dst)
+                )
+
+    def package_info(self):
+        if self.options.shared:
+            self.cpp_info.libs = ["onnxruntime"]
+        else:
+            debug_suffix = "d" if self.settings.build_type == "Debug" else ""
+
+            onnxruntime_libs = []
+
+            if self.options.with_nnapi:
+                onnxruntime_libs.append("providers_nnapi")
+
+            onnxruntime_libs.extend([
+                "session",
+                "optimizer",
+                "providers",
+                "framework",
+                "graph",
+                "util",
+                "mlas",
+                "common",
+                "flatbuffers",
+            ])
+
+            self.cpp_info.libs = \
+                [f"onnxruntime_{lib}" for lib in onnxruntime_libs]
+
+            self.cpp_info.libs.extend([
+                "onnx",
+                "onnx_proto",
+                "flatbuffers",
+                "re2",
+                "nsync_cpp",
+                "protobuf-lite" + debug_suffix,
+            ])
+
+            if self.options.with_dnnl:
+                self.cpp_info.libs.append("dnnl")
+
+        self.cpp_info.includedirs.append("include/onnxruntime/core/session")
+
+    @property
+    def _onnxruntime_build_dir(self):
+        msr = self.options.force_min_size_rel
+        build_type = "MinSizeRel" if msr else str(self.settings.build_type)
+        return os.path.join(
+            self.build_folder,
+            self._source_subfolder,
+            "build",
+            str(self.settings.os),
+            build_type
+        )
diff --git a/recipes/onnxruntime/all/test_package/CMakeLists.txt b/recipes/onnxruntime/all/test_package/CMakeLists.txt
new file mode 100644
index 00000000000000..01078624049d16
--- /dev/null
+++ b/recipes/onnxruntime/all/test_package/CMakeLists.txt
@@ -0,0 +1,14 @@
+cmake_minimum_required(VERSION 3.1)
+project(test_package)
+
+include(${CMAKE_BINARY_DIR}/conanbuildinfo.cmake)
+conan_basic_setup(TARGETS)
+
+find_package(OnnxRuntime REQUIRED)
+
+set(CMAKE_CXX_STANDARD 11)
+set(CMAKE_CXX_STANDARD_REQUIRED ON)
+
+add_executable(${PROJECT_NAME} test_package.cpp)
+target_link_libraries(${PROJECT_NAME} ${OnnxRuntime_LIBRARIES})
+target_include_directories(${PROJECT_NAME} PRIVATE ${OnnxRuntime_INCLUDE_DIRS})
diff --git a/recipes/onnxruntime/all/test_package/conanfile.py b/recipes/onnxruntime/all/test_package/conanfile.py
new file mode 100644
index 00000000000000..20a73d2658fc51
--- /dev/null
+++ b/recipes/onnxruntime/all/test_package/conanfile.py
@@ -0,0 +1,23 @@
+from conans import ConanFile, CMake, tools
+import os
+
+
+class OnnxRuntimeTestConan(ConanFile):
+    settings = "os", "compiler", "build_type", "arch"
+    generators = "cmake", "cmake_find_package"
+
+    def build(self):
+        cmake = CMake(self)
+        cmake.configure()
+        cmake.build()
+
+    def test(self):
+        tools.download(
+            "https://github.com/microsoft/onnxruntime/raw"
+            "/master/csharp/testdata/squeezenet.onnx",
+            "squeezenet.onnx"
+        )
+
+        if not tools.cross_building(self.settings):
+            bin_path = os.path.join("bin", "test_package")
+            self.run(bin_path, run_environment=True)
diff --git a/recipes/onnxruntime/all/test_package/test_package.cpp b/recipes/onnxruntime/all/test_package/test_package.cpp
new file mode 100644
index 00000000000000..aefed9b0a6b0e8
--- /dev/null
+++ b/recipes/onnxruntime/all/test_package/test_package.cpp
@@ -0,0 +1,132 @@
+// Copyright(c) Microsoft Corporation.All rights reserved.
+// Licensed under the MIT License.
+//
+// Source: https://github.com/microsoft/onnxruntime/blob/master/csharp/test/Microsoft.ML.OnnxRuntime.EndToEndTests.Capi/CXX_Api_Sample.cpp
+
+#include <assert.h>
+#include <vector>
+#include <onnxruntime_cxx_api.h>
+
+int main(int argc, char* argv[]) {
+  //*************************************************************************
+  // initialize  enviroment...one enviroment per process
+  // enviroment maintains thread pools and other state info
+  Ort::Env env(ORT_LOGGING_LEVEL_WARNING, "test");
+
+  // initialize session options if needed
+  Ort::SessionOptions session_options;
+  session_options.SetIntraOpNumThreads(1);
+
+  // If onnxruntime.dll is built with CUDA enabled, we can uncomment out this line to use CUDA for this
+  // session (we also need to include cuda_provider_factory.h above which defines it)
+  // #include "cuda_provider_factory.h"
+  // OrtSessionOptionsAppendExecutionProvider_CUDA(session_options, 1);
+
+  // Sets graph optimization level
+  // Available levels are
+  // ORT_DISABLE_ALL -> To disable all optimizations
+  // ORT_ENABLE_BASIC -> To enable basic optimizations (Such as redundant node removals)
+  // ORT_ENABLE_EXTENDED -> To enable extended optimizations (Includes level 1 + more complex optimizations like node fusions)
+  // ORT_ENABLE_ALL -> To Enable All possible opitmizations
+  session_options.SetGraphOptimizationLevel(GraphOptimizationLevel::ORT_ENABLE_EXTENDED);
+
+  //*************************************************************************
+  // create session and load model into memory
+  // using squeezenet version 1.3
+  // URL = https://github.com/onnx/models/tree/master/squeezenet
+#ifdef _WIN32
+  const wchar_t* model_path = L"squeezenet.onnx";
+#else
+  const char* model_path = "squeezenet.onnx";
+#endif
+
+  printf("Using Onnxruntime C++ API\n");
+  Ort::Session session(env, model_path, session_options);
+
+  //*************************************************************************
+  // print model input layer (node names, types, shape etc.)
+  Ort::AllocatorWithDefaultOptions allocator;
+
+  // print number of model input nodes
+  size_t num_input_nodes = session.GetInputCount();
+  std::vector<const char*> input_node_names(num_input_nodes);
+  std::vector<int64_t> input_node_dims;  // simplify... this model has only 1 input node {1, 3, 224, 224}.
+                                         // Otherwise need vector<vector<>>
+
+  printf("Number of inputs = %zu\n", num_input_nodes);
+
+  // iterate over all input nodes
+  for (int i = 0; i < num_input_nodes; i++) {
+    // print input node names
+    char* input_name = session.GetInputName(i, allocator);
+    printf("Input %d : name=%s\n", i, input_name);
+    input_node_names[i] = input_name;
+
+    // print input node types
+    Ort::TypeInfo type_info = session.GetInputTypeInfo(i);
+    auto tensor_info = type_info.GetTensorTypeAndShapeInfo();
+
+    ONNXTensorElementDataType type = tensor_info.GetElementType();
+    printf("Input %d : type=%d\n", i, type);
+
+    // print input shapes/dims
+    input_node_dims = tensor_info.GetShape();
+    printf("Input %d : num_dims=%zu\n", i, input_node_dims.size());
+    for (int j = 0; j < input_node_dims.size(); j++)
+      printf("Input %d : dim %d=%lld\n", i, j, input_node_dims[j]);
+  }
+
+  // Results should be...
+  // Number of inputs = 1
+  // Input 0 : name = data_0
+  // Input 0 : type = 1
+  // Input 0 : num_dims = 4
+  // Input 0 : dim 0 = 1
+  // Input 0 : dim 1 = 3
+  // Input 0 : dim 2 = 224
+  // Input 0 : dim 3 = 224
+
+  //*************************************************************************
+  // Similar operations to get output node information.
+  // Use OrtSessionGetOutputCount(), OrtSessionGetOutputName()
+  // OrtSessionGetOutputTypeInfo() as shown above.
+
+  //*************************************************************************
+  // Score the model using sample data, and inspect values
+
+  size_t input_tensor_size = 224 * 224 * 3;  // simplify ... using known dim values to calculate size
+                                             // use OrtGetTensorShapeElementCount() to get official size!
+
+  std::vector<float> input_tensor_values(input_tensor_size);
+  std::vector<const char*> output_node_names = {"softmaxout_1"};
+
+  // initialize input data with values in [0.0, 1.0]
+  for (unsigned int i = 0; i < input_tensor_size; i++)
+    input_tensor_values[i] = (float)i / (input_tensor_size + 1);
+
+  // create input tensor object from data values
+  auto memory_info = Ort::MemoryInfo::CreateCpu(OrtArenaAllocator, OrtMemTypeDefault);
+  Ort::Value input_tensor = Ort::Value::CreateTensor<float>(memory_info, input_tensor_values.data(), input_tensor_size, input_node_dims.data(), 4);
+  assert(input_tensor.IsTensor());
+
+  // score model & input tensor, get back output tensor
+  auto output_tensors = session.Run(Ort::RunOptions{nullptr}, input_node_names.data(), &input_tensor, 1, output_node_names.data(), 1);
+  assert(output_tensors.size() == 1 && output_tensors.front().IsTensor());
+
+  // Get pointer to output tensor float values
+  float* floatarr = output_tensors.front().GetTensorMutableData<float>();
+  assert(abs(floatarr[0] - 0.000045) < 1e-6);
+
+  // score the model, and print scores for first 5 classes
+  for (int i = 0; i < 5; i++)
+    printf("Score for class [%d] =  %f\n", i, floatarr[i]);
+
+  // Results should be as below...
+  // Score for class[0] = 0.000045
+  // Score for class[1] = 0.003846
+  // Score for class[2] = 0.000125
+  // Score for class[3] = 0.001180
+  // Score for class[4] = 0.001317
+  printf("Done!\n");
+  return 0;
+}
diff --git a/recipes/onnxruntime/config.yml b/recipes/onnxruntime/config.yml
new file mode 100644
index 00000000000000..4c7c778dfa5404
--- /dev/null
+++ b/recipes/onnxruntime/config.yml
@@ -0,0 +1,7 @@
+versions:
+  "1.7.1":
+    folder: all
+  "1.5.3":
+    folder: all
+  "1.2.0":
+    folder: all