diff --git a/mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp b/mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp
--- a/mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp
+++ b/mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp
@@ -786,6 +786,59 @@
   return success();
 }
 
+llvm::AtomicOrdering convertAtomicOrdering(Optional<StringRef> AOAttr) {
+  if (!AOAttr.hasValue())
+    return llvm::AtomicOrdering::Monotonic; // Default Memory Ordering
+
+  llvm::AtomicOrdering AO = llvm::AtomicOrdering::Monotonic;
+  StringRef AOString = AOAttr.getValue();
+  if (AOString == "seq_cst")
+    AO = llvm::AtomicOrdering::SequentiallyConsistent;
+  else if (AOString == "acq_rel")
+    AO = llvm::AtomicOrdering::AcquireRelease;
+  else if (AOString == "acquire")
+    AO = llvm::AtomicOrdering::Acquire;
+  else if (AOString == "release")
+    AO = llvm::AtomicOrdering::Release;
+  else if (AOString == "relaxed")
+    AO = llvm::AtomicOrdering::Monotonic;
+  return AO;
+}
+
+static LogicalResult
+convertOmpAtomicRead(Operation &opInst, llvm::IRBuilderBase &builder,
+                     LLVM::ModuleTranslation &moduleTranslation) {
+
+  auto readOp = cast<omp::AtomicReadOp>(opInst);
+  llvm::OpenMPIRBuilder *ompBuilder = moduleTranslation.getOpenMPBuilder();
+
+  // Set up the source location value for OpenMP runtime.
+  llvm::DISubprogram *subprogram =
+      builder.GetInsertBlock()->getParent()->getSubprogram();
+  const llvm::DILocation *diLoc =
+      moduleTranslation.translateLoc(opInst.getLoc(), subprogram);
+  llvm::OpenMPIRBuilder::LocationDescription ompLoc(builder.saveIP(),
+                                                    llvm::DebugLoc(diLoc));
+  llvm::AtomicOrdering AO = convertAtomicOrdering(readOp.memory_order());
+  llvm::Value *address = moduleTranslation.lookupValue(readOp.address());
+  llvm::OpenMPIRBuilder::InsertPointTy currentIP = builder.saveIP();
+
+  // Insert alloca for result
+  llvm::OpenMPIRBuilder::InsertPointTy allocaIP =
+      findAllocaInsertPoint(builder, moduleTranslation);
+  builder.restoreIP(allocaIP);
+  llvm::Value *v = builder.CreateAlloca(
+      moduleTranslation.convertType(readOp.getResult().getType()));
+  moduleTranslation.mapValue(readOp.getResult(), v);
+
+  // Restore the IP and insert Atomic Read
+  builder.restoreIP(currentIP);
+  llvm::OpenMPIRBuilder::AtomicOpValue V = {v, false, false};
+  llvm::OpenMPIRBuilder::AtomicOpValue X = {address, false, false};
+  builder.restoreIP(ompBuilder->createAtomicRead(ompLoc, X, V, AO));
+  return success();
+}
+
 /// Converts an OpenMP reduction operation using OpenMPIRBuilder. Expects the
 /// mapping between reduction variables and their private equivalents to have
 /// been stored on the ModuleTranslation stack. Currently only supports
@@ -907,6 +960,9 @@
       .Case([&](omp::WsLoopOp) {
         return convertOmpWsLoop(*op, builder, moduleTranslation);
       })
+      .Case([&](omp::AtomicReadOp) {
+        return convertOmpAtomicRead(*op, builder, moduleTranslation);
+      })
       .Case<omp::YieldOp, omp::TerminatorOp, omp::ReductionDeclareOp,
             omp::CriticalDeclareOp>([](auto op) {
         // `yield` and `terminator` can be just omitted. The block structure
diff --git a/mlir/test/Target/LLVMIR/openmp-llvm.mlir b/mlir/test/Target/LLVMIR/openmp-llvm.mlir
--- a/mlir/test/Target/LLVMIR/openmp-llvm.mlir
+++ b/mlir/test/Target/LLVMIR/openmp-llvm.mlir
@@ -706,3 +706,33 @@
 
   llvm.return
 }
+
+// -----
+
+// CHECK-LABEL: @omp_atomic_read
+// CHECK-SAME: (i32* %[[ARG0:.*]])
+llvm.func @omp_atomic_read(%arg0 : !llvm.ptr<i32>) -> () {
+  // CHECK: %{{.*}} = alloca i32, align 4
+  // CHECK: %{{.*}} = alloca i32, align 4
+  // CHECK: %{{.*}} = alloca i32, align 4
+  // CHECK: %{{.*}} = alloca i32, align 4
+
+  // CHECK: %[[X1:.*]] = load atomic i32, i32* %[[ARG0]] monotonic, align 4
+  // CHECK: store i32 %[[X1]], i32* %{{.*}}, align 4
+  %x1 = omp.atomic.read %arg0 : !llvm.ptr<i32> -> i32
+
+  // CHECK: %[[X2:.*]] = load atomic i32, i32* %[[ARG0]] seq_cst, align 4
+  // CHECK: call void @__kmpc_flush(%{{.*}})
+  // CHECK: store i32 %[[X2]], i32* %{{.*}}, align 4
+  %x2 = omp.atomic.read %arg0 memory_order(seq_cst) : !llvm.ptr<i32> -> i32
+
+  // CHECK: %[[X3:.*]] = load atomic i32, i32* %[[ARG0]] acquire, align 4
+  // CHECK: call void @__kmpc_flush(%{{.*}})
+  // CHECK: store i32 %[[X3]], i32* %{{.*}}, align 4
+  %x3 = omp.atomic.read %arg0 memory_order(acquire) : !llvm.ptr<i32> -> i32
+
+  // CHECK: %[[X4:.*]] = load atomic i32, i32* %[[ARG0]] monotonic, align 4
+  // CHECK: store i32 %[[X4]], i32* %{{.*}}, align 4
+  %x4 = omp.atomic.read %arg0 memory_order(relaxed) : !llvm.ptr<i32> -> i32
+  llvm.return
+}
\ No newline at end of file