diff --git a/tools/src/CSVFileImport.cc b/tools/src/CSVFileImport.cc
index 7bbc9c64ae..393367babe 100644
--- a/tools/src/CSVFileImport.cc
+++ b/tools/src/CSVFileImport.cc
@@ -23,6 +23,7 @@
 #include <algorithm>
 #include <fstream>
 #include <iostream>
+#include <list>
 #include <memory>
 #include <getopt.h>
 #include <string>
@@ -77,8 +78,8 @@ void fillStringValues(const std::vector<std::string>& data,
                       orc::ColumnVectorBatch* batch,
                       uint64_t numValues,
                       uint64_t colIndex,
-                      orc::DataBuffer<char>& buffer,
-                      uint64_t& offset) {
+                      orc::DataBuffer<char>& buffer) {
+  uint64_t offset = 0;
   orc::StringVectorBatch* stringBatch =
     dynamic_cast<orc::StringVectorBatch*>(batch);
   bool hasNull = false;
@@ -365,7 +366,8 @@ int main(int argc, char* argv[]) {
   double totalElapsedTime = 0.0;
   clock_t totalCPUTime = 0;
 
-  orc::DataBuffer<char> buffer(*orc::getDefaultPool(), 4 * 1024 * 1024);
+  typedef std::list<orc::DataBuffer<char>> DataBufferList;
+  DataBufferList bufferList;
 
   orc::WriterOptions options;
   options.setStripeSize(stripeSize);
@@ -385,7 +387,6 @@ int main(int argc, char* argv[]) {
   std::ifstream finput(input.c_str());
   while (!eof) {
     uint64_t numValues = 0;      // num of lines read in a batch
-    uint64_t bufferOffset = 0;   // current offset in the string buffer
 
     data.clear();
     memset(rowBatch->notNull.data(), 1, batchSize);
@@ -420,13 +421,13 @@ int main(int argc, char* argv[]) {
           case orc::STRING:
           case orc::CHAR:
           case orc::VARCHAR:
-          case orc::BINARY:
+          case orc::BINARY: 
+            bufferList.emplace_back(*orc::getDefaultPool(), 1 * 1024 * 1024);
             fillStringValues(data,
                              structBatch->fields[i],
                              numValues,
                              i,
-                             buffer,
-                             bufferOffset);
+                             bufferList.back());
             break;
           case orc::FLOAT:
           case orc::DOUBLE:
diff --git a/tools/test/TestCSVFileImport.cc b/tools/test/TestCSVFileImport.cc
index b4cd9679f1..0ae7f2cf1f 100644
--- a/tools/test/TestCSVFileImport.cc
+++ b/tools/test/TestCSVFileImport.cc
@@ -24,6 +24,8 @@
 #include "wrap/gmock.h"
 #include "wrap/gtest-wrapper.h"
 
+#include <fstream>
+
 TEST (TestCSVFileImport, test10rows) {
   // create an ORC file from importing the CSV file
   const std::string pgm1 = findProgram("tools/src/csv-import");
@@ -90,3 +92,36 @@ TEST (TestCSVFileImport, testTimezoneOption) {
     EXPECT_EQ("", error);
   }
 }
+
+TEST (TestCSVFileImport, testLongString) {
+  // create an ORC file from importing the CSV file
+  const std::string pgm1 = findProgram("tools/src/csv-import");
+  const std::string csvFile = "/tmp/test_csv_import_test_long_string.csv";
+  const std::string orcFile = "/tmp/test_csv_import_test_long_string.orc";
+  const std::string schema = "'struct<_a:string,b_:binary,_c:varchar(10)>'";
+  std::string output;
+  std::string error;
+
+  std::ofstream csvFileStream(csvFile, std::ios::binary | std::ios::out | std::ios::trunc);
+  if(csvFileStream.is_open())
+  {
+    std::string longStr;
+    longStr.resize(4 * 1024 * 1024 + 1, 'x');
+    csvFileStream << "str1," << longStr << ",var1\n";
+    csvFileStream << "str2," << longStr << ",var2\n";
+    csvFileStream.close();
+  }
+
+  EXPECT_EQ(0, runProgram({pgm1, schema, csvFile, orcFile}, output, error));
+  EXPECT_EQ("", error);
+
+  // verify the ORC file content
+  const std::string pgm2 = findProgram("tools/src/orc-contents");
+  std::string option = "--columns=0,2";
+  const std::string expected =
+    "{\"_a\": \"str1\", \"_c\": \"var1\"}\n"
+    "{\"_a\": \"str2\", \"_c\": \"var2\"}\n";
+  EXPECT_EQ(0, runProgram({pgm2, option, orcFile}, output, error));
+  EXPECT_EQ(expected, output);
+  EXPECT_EQ("", error);
+}