diff --git a/CMakeLists.txt b/CMakeLists.txt
index b297423..162b509 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -516,9 +516,10 @@
              src/armnnTfLiteParser/test/ResizeBilinear.cpp
              src/armnnTfLiteParser/test/Softmax.cpp
              src/armnnTfLiteParser/test/SpaceToBatchND.cpp
-             src/armnnTfLiteParser/test/Sub.cpp
+             src/armnnTfLiteParser/test/Split.cpp
              src/armnnTfLiteParser/test/Squeeze.cpp
              src/armnnTfLiteParser/test/StridedSlice.cpp
+             src/armnnTfLiteParser/test/Sub.cpp
              src/armnnTfLiteParser/test/LoadModel.cpp
              src/armnnTfLiteParser/test/GetBuffer.cpp
              src/armnnTfLiteParser/test/OutputShapeOfSqueeze.cpp
diff --git a/src/armnnTfLiteParser/TensorFlowLiteSupport.md b/src/armnnTfLiteParser/TensorFlowLiteSupport.md
index 84734c5..afbe2ce 100644
--- a/src/armnnTfLiteParser/TensorFlowLiteSupport.md
+++ b/src/armnnTfLiteParser/TensorFlowLiteSupport.md
@@ -48,6 +48,8 @@
 
 * SPACE_TO_BATCH
 
+* SPLIT
+
 * SQUEEZE
 
 * STRIDED_SLICE
diff --git a/src/armnnTfLiteParser/TfLiteParser.cpp b/src/armnnTfLiteParser/TfLiteParser.cpp
index b9a3522..c00c218 100644
--- a/src/armnnTfLiteParser/TfLiteParser.cpp
+++ b/src/armnnTfLiteParser/TfLiteParser.cpp
@@ -464,6 +464,7 @@
     m_ParserFunctions[tflite::BuiltinOperator_MUL]               =  &TfLiteParser::ParseMul;
     m_ParserFunctions[tflite::BuiltinOperator_MEAN]              =  &TfLiteParser::ParseMean;
     m_ParserFunctions[tflite::BuiltinOperator_PAD]               =  &TfLiteParser::ParsePad;
+    m_ParserFunctions[tflite::BuiltinOperator_SPLIT]             =  &TfLiteParser::ParseSplit;
 }
 
 void TfLiteParser::ResetParser()
@@ -1851,6 +1852,94 @@
                                                               outputTensorIndexes[3]});
 }
 
+void TfLiteParser::ParseSplit(size_t subgraphIndex, size_t operatorIndex)
+{
+    CHECK_MODEL(m_Model, subgraphIndex, operatorIndex);
+
+    const auto & operatorPtr = m_Model->subgraphs[subgraphIndex]->operators[operatorIndex];
+    const auto * options = operatorPtr->builtin_options.AsSplitOptions();
+
+    const unsigned int numSplits = CHECKED_NON_NEGATIVE(options->num_splits);
+
+    auto inputs = GetInputs(m_Model, subgraphIndex, operatorIndex);
+    CHECK_VALID_SIZE(inputs.size(), 2);
+    auto outputs = GetOutputs(m_Model, subgraphIndex, operatorIndex);
+    CHECK_VALID_SIZE(outputs.size(), numSplits);
+
+    armnn::TensorInfo inputTensorInfo  = ToTensorInfo(inputs[0]);
+    armnn::TensorInfo axisTensorInfo = ToTensorInfo(inputs[1]);
+
+    // This splitDim indicates the data format: 3 is the NHWC, 1 is the NCHW.
+    const unsigned int splitDim = static_cast<unsigned int>(axisTensorInfo.GetShape()[0]);
+
+    // Armnn supports split along the channel dimension for data formats NHWC and NCHW.
+    if (splitDim == 0 || splitDim == 2)
+    {
+        throw ParseException(
+            boost::str(
+                boost::format(
+                    "Dimension %1% for split is not supported by Armnn. %2%")
+                    % splitDim
+                    % CHECK_LOCATION().AsString()));
+    }
+
+    auto inputDimSize = inputTensorInfo.GetNumDimensions();
+    if (inputDimSize != MaxNumOfTensorDimensions)
+    {
+        throw ParseException(
+            boost::str(
+                boost::format(
+                    "The number of dimensions: %1% for input tensors of the "
+                    "split op should be %2% %3%")
+                    % inputTensorInfo.GetNumDimensions()
+                    % MaxNumOfTensorDimensions
+                    % CHECK_LOCATION().AsString()));
+    }
+
+    std::vector<unsigned int> splitterDimSizes(inputDimSize);
+
+    // Add current input shape to splitterDimSizes
+    for (unsigned int i = 0; i < inputDimSize; ++i)
+    {
+        splitterDimSizes[i] = inputTensorInfo.GetShape()[i];
+    }
+
+    if (splitterDimSizes[splitDim] % numSplits != 0)
+    {
+        throw ParseException("Number of splits must evenly divide the dimension");
+    }
+    splitterDimSizes[splitDim] /= numSplits;
+
+    SplitterDescriptor splitDesc(numSplits);
+    for (unsigned int j = 0; j < numSplits; ++j)
+    {
+        // Set the size of the views.
+        for (unsigned int dimIdx = 0; dimIdx < splitterDimSizes.size(); ++dimIdx)
+        {
+            splitDesc.SetViewSize(j, dimIdx, splitterDimSizes[dimIdx]);
+        }
+        splitDesc.SetViewOriginCoord(j, splitDim, splitterDimSizes[splitDim] * j);
+    }
+
+    auto layerName = boost::str(boost::format("Split:%1%:%2%") % subgraphIndex % operatorIndex);
+    IConnectableLayer* layer = m_Network->AddSplitterLayer(splitDesc, layerName.c_str());
+
+    auto inputTensorIndexes = AsUnsignedVector(GetInputTensorIds(m_Model, subgraphIndex, operatorIndex));
+    RegisterInputSlots(subgraphIndex, operatorIndex, layer, {inputTensorIndexes[0]});
+
+    TensorShape outShape = TensorShape(static_cast<unsigned int>(splitterDimSizes.size()),
+                                       splitterDimSizes.data());
+
+    for (unsigned int k = 0; k < layer->GetNumOutputSlots(); ++k)
+    {
+        layer->GetOutputSlot(k).SetTensorInfo(armnn::TensorInfo(outShape,
+                                                                inputTensorInfo.GetDataType()));
+    }
+
+    auto outputTensorIndexes = AsUnsignedVector(GetOutputTensorIds(m_Model, subgraphIndex, operatorIndex));
+    RegisterOutputSlots(subgraphIndex, operatorIndex, layer, outputTensorIndexes);
+}
+
 armnn::IConnectableLayer* TfLiteParser::AddFusedActivationLayer(armnn::IConnectableLayer* prevLayer,
                                                                 unsigned int outputSlot,
                                                                 tflite::ActivationFunctionType activationType)
diff --git a/src/armnnTfLiteParser/TfLiteParser.hpp b/src/armnnTfLiteParser/TfLiteParser.hpp
index 2895487..e166dd5 100644
--- a/src/armnnTfLiteParser/TfLiteParser.hpp
+++ b/src/armnnTfLiteParser/TfLiteParser.hpp
@@ -115,8 +115,8 @@
     void ParseMul(size_t subgraphIndex, size_t operatorIndex);
     void ParseMean(size_t subgraphIndex, size_t operatorIndex);
     void ParsePad(size_t subgraphIndex, size_t operatorIndex);
-
     void ParsePool(size_t subgraphIndex, size_t operatorIndex, armnn::PoolingAlgorithm algorithm);
+    void ParseSplit(size_t subgraphIndex, size_t operatorIndex);
 
     void RegisterProducerOfTensor(size_t subgraphIndex, size_t tensorIndex, armnn::IOutputSlot* slot);
     void RegisterConsumerOfTensor(size_t subgraphIndex, size_t tensorIndex, armnn::IInputSlot* slot);
diff --git a/src/armnnTfLiteParser/test/Split.cpp b/src/armnnTfLiteParser/test/Split.cpp
new file mode 100644
index 0000000..774a416
--- /dev/null
+++ b/src/armnnTfLiteParser/test/Split.cpp
@@ -0,0 +1,137 @@
+//
+// Copyright © 2017 Arm Ltd. All rights reserved.
+// SPDX-License-Identifier: MIT
+//
+
+#include <boost/test/unit_test.hpp>
+#include "ParserFlatbuffersFixture.hpp"
+#include "../TfLiteParser.hpp"
+
+#include <string>
+#include <iostream>
+
+BOOST_AUTO_TEST_SUITE(TensorflowLiteParser)
+
+struct SplitFixture : public ParserFlatbuffersFixture
+{
+    explicit SplitFixture(const std::string & inputShape,
+                          const std::string & axisShape,
+                          const std::string & numSplits,
+                          const std::string & outputShape1,
+                          const std::string & outputShape2)
+    {
+        m_JsonString = R"(
+            {
+                "version": 3,
+                "operator_codes": [ { "builtin_code": "SPLIT" } ],
+                "subgraphs": [ {
+                    "tensors": [
+                        {
+                            "shape": )" + inputShape + R"(,
+                            "type": "FLOAT32",
+                            "buffer": 0,
+                            "name": "inputTensor",
+                            "quantization": {
+                                "min": [ 0.0 ],
+                                "max": [ 255.0 ],
+                                "scale": [ 1.0 ],
+                                "zero_point": [ 0 ],
+                            }
+                        },
+                        {
+                            "shape": )" + axisShape + R"(,
+                            "type": "INT32",
+                            "buffer": 1,
+                            "name": "axis",
+                            "quantization": {
+                                "min": [ 0.0 ],
+                                "max": [ 255.0 ],
+                                "scale": [ 1.0 ],
+                                "zero_point": [ 0 ],
+                            }
+                        },
+                        {
+                            "shape": )" + outputShape1 + R"( ,
+                            "type": "FLOAT32",
+                            "buffer": 2,
+                            "name": "outputTensor1",
+                            "quantization": {
+                                "min": [ 0.0 ],
+                                "max": [ 255.0 ],
+                                "scale": [ 1.0 ],
+                                "zero_point": [ 0 ],
+                            }
+                        },
+                        {
+                            "shape": )" + outputShape2 + R"( ,
+                            "type": "FLOAT32",
+                            "buffer": 3,
+                            "name": "outputTensor2",
+                            "quantization": {
+                                "min": [ 0.0 ],
+                                "max": [ 255.0 ],
+                                "scale": [ 1.0 ],
+                                "zero_point": [ 0 ],
+                            }
+                        }
+                    ],
+                    "inputs": [ 0, 1 ],
+                    "outputs": [ 2, 3 ],
+                    "operators": [
+                        {
+                            "opcode_index": 0,
+                            "inputs": [ 0, 1 ],
+                            "outputs": [ 2, 3 ],
+                            "builtin_options_type": "SplitOptions",
+                            "builtin_options": {
+                                "num_splits": )" + numSplits + R"(
+                            },
+                            "custom_options_format": "FLEXBUFFERS"
+                        }
+                    ],
+                } ],
+                "buffers" : [ {}, {} ]
+            }
+        )";
+
+        Setup();
+    }
+};
+
+
+struct SimpleSplitFixture : SplitFixture
+{
+    SimpleSplitFixture() : SplitFixture( "[ 2, 2, 2, 2 ]", "[ 1 ]", "2",
+        "[ 2, 1, 2, 2 ]", "[ 2, 1, 2, 2 ]")
+         {}
+};
+
+BOOST_FIXTURE_TEST_CASE(ParseAxisOneSplitTwo, SimpleSplitFixture)
+{
+
+    RunTest<4, armnn::DataType::Float32>(
+        0,
+        { {"inputTensor", { 1.0f, 2.0f, 3.0f, 4.0f, 5.0f, 6.0f, 7.0f, 8.0f, 9.0f, 10.0f,
+                            11.0f, 12.0f, 13.0f, 14.0f, 15.0f, 16.0f } } },
+        { {"outputTensor1", { 1.0f, 2.0f, 3.0f, 4.0f, 9.0f, 10.0f, 11.0f, 12.0f }},
+          {"outputTensor2", { 5.0f, 6.0f, 7.0f, 8.0f, 13.0f, 14.0f, 15.0f, 16.0f }}});
+}
+
+struct SimpleSplitAxisThreeFixture : SplitFixture
+{
+    SimpleSplitAxisThreeFixture() : SplitFixture( "[ 2, 2, 2, 2 ]", "[ 3 ]", "2",
+        "[ 2, 2, 2, 1 ]", "[ 2, 2, 2, 1 ]")
+    {}
+};
+
+BOOST_FIXTURE_TEST_CASE(ParseAxisThreeSplitTwo, SimpleSplitAxisThreeFixture)
+{
+    RunTest<4, armnn::DataType::Float32>(
+        0,
+        { {"inputTensor", { 1.0f, 2.0f, 3.0f, 4.0f, 5.0f, 6.0f, 7.0f, 8.0f, 9.0f, 10.0f,
+                            11.0f, 12.0f, 13.0f, 14.0f, 15.0f, 16.0f } } },
+        { {"outputTensor1", { 1.0f, 3.0f, 5.0f, 7.0f, 9.0f, 11.0f, 13.0f, 15.0f }},
+          {"outputTensor2", { 2.0f, 4.0f, 6.0f, 8.0f, 10.0f, 12.0f, 14.0f, 16.0f } } } );
+}
+
+BOOST_AUTO_TEST_SUITE_END()
\ No newline at end of file