Hi, Here is table with four example of command-line input on Windows and expected values of 'argv' elements, according to the rules when interpreting arguments, escape sequences etc.: https://msdn.microsoft.com/en-us/library/17w5ykft.aspx Double-checking this actually works: ///////// Windows console program ////////////////////////////////// #include <cstdio> int main(int argc, char* argv[]) { for (int i = 0; i < argc; ++i) std::printf("%d: %s\n", i, argv[i]); } F:\test_cpp2015\Debug>test_cpp2015.exe "abc" d e 0: test_cpp2015.exe 1: abc 2: d 3: e F:\test_cpp2015\Debug>test_cpp2015.exe a\\b d"e f"g h 0: test_cpp2015.exe 1: a\\b 2: de fg 3: h F:\test_cpp2015\Debug>test_cpp2015.exe a\\\"b c d 0: test_cpp2015.exe 1: a\"b 2: c 3: d F:\test_cpp2015\Debug>test_cpp2015.exe a\\\\"b c" d e 0: test_cpp2015.exe 1: a\\b c 2: d 3: e ////////////////////////////////// Now, the same test but with values of framework::master_test_suite().argv ///////// Boost.Test program /////////////////////////////////////// #define BOOST_TEST_MAIN #include <boost/test/unit_test.hpp> BOOST_AUTO_TEST_CASE(argv_parsing) { auto argc = boost::unit_test::framework::master_test_suite().argc; auto argv = boost::unit_test::framework::master_test_suite().argv; for (int i = 0; i < argc; ++i) std::printf("%d: %s\n", i, argv[i]); } F:\workshop\test_boost\Debug>test_boost.exe "abc" d e Running 1 test case... 0: test_boost.exe 1: abc 2: d 3: e F:\workshop\test_boost\Debug>test_boost.exe a\\b d"e f"g h Running 1 test case... 0: test_boost.exe 1: a\\b 2: de 3: fg 4: h F:\workshop\test_boost\Debug>test_boost.exe a\\\"b c d Running 1 test case... 0: test_boost.exe 1: a\"b 2: c 3: d F:\workshop\test_boost\Debug>test_boost.exe a\\\\"b c" d e Running 1 test case... 0: test_boost.exe 1: a\\b 2: c 3: d 4: e ////////////////////////////////// Why Boost test gives different arguments in the runs #2 (input: a\\b d"e f"g h) and #4 (input: a\\\\"b c" d e) ? I can't see any caveats mentioned in the documentation http://www.boost.org/doc/libs/1_59_0/libs/test/doc/html/boost_test/tests_org... Best regards, -- Mateusz Loskot, http://mateusz.loskot.net