diff --git a/NEWS b/NEWS
index e4d97c321..3ea2ded9d 100644
--- a/NEWS
+++ b/NEWS
@@ -3,6 +3,14 @@ For full details, see the git log at: https://github.com/ksh93/ksh
 
 Any uppercase BUG_* names are modernish shell bug IDs.
 
+2020-07-25:
+
+- Fixed BUG_MULTIBIFS: Multibyte characters can now be used as IFS
+  delimiters. "$*" was incorrectly joining positional parameters on
+  the first byte of a multibyte character. This was due to truncation
+  based on the incorrect assumption the IFS would never be larger
+  than a single byte.
+
 2020-07-23:
 
 - Fixed an infinite loop that could occur when ksh is the system's /bin/sh.
diff --git a/TODO b/TODO
index f1e7aa0e7..36026fac2 100644
--- a/TODO
+++ b/TODO
@@ -54,9 +54,3 @@ https://github.com/modernish/modernish/tree/0.16/lib/modernish/cap/
   between 'while'/'until' and 'do'), the exit status passed down from the
   previous command is ignored and the function returns with status 0
   instead.
-
-- BUG_MULTIBIFS: We're on a UTF-8 locale and the shell supports UTF-8
-  characters in general (i.e. we don't have WRN_MULTIBYTE) – however, using
-  multi-byte characters as IFS field delimiters still doesn't work. For
-  example, "$*" joins positional parameters on the first byte of IFS instead
-  of the first character.
diff --git a/src/cmd/ksh93/include/version.h b/src/cmd/ksh93/include/version.h
index 2f87d7609..066f85373 100644
--- a/src/cmd/ksh93/include/version.h
+++ b/src/cmd/ksh93/include/version.h
@@ -17,4 +17,4 @@
 *                  David Korn <dgk@research.att.com>                   *
 *                                                                      *
 ***********************************************************************/
-#define SH_RELEASE	"93u+m 2020-07-23"
+#define SH_RELEASE	"93u+m 2020-07-25"
diff --git a/src/cmd/ksh93/sh/macro.c b/src/cmd/ksh93/sh/macro.c
index 530da993c..1fb67ffb3 100644
--- a/src/cmd/ksh93/sh/macro.c
+++ b/src/cmd/ksh93/sh/macro.c
@@ -1954,10 +1954,34 @@ retry2:
 			}
 			else if(d)
 			{
+#if SHOPT_MULTIBYTE
+				Sfio_t *sfio_ptr = (mp->sp) ? mp->sp : stkp;
+
+				/*
+				 * We know from above that if we are not performing @-expansion
+				 * then we assigned `d` the value of `mp->ifs`, here we check
+				 * whether or not we have a valid string of IFS characters to
+				 * write as it is possible for `d` to be set to `mp->ifs` and
+				 * yet `mp->ifsp` to be NULL.
+				 */
+				if(mode != '@' && mp->ifsp)
+				{
+					/*
+					 * Handle multi-byte characters being used for the internal
+					 * field separator (IFS).
+					 */
+					int i;
+					for(i = 0; i < mbsize(mp->ifsp); i++)
+						sfputc(sfio_ptr,mp->ifsp[i]);
+				}
+				else
+					sfputc(sfio_ptr,d);
+#else
 				if(mp->sp)
 					sfputc(mp->sp,d);
 				else
 					sfputc(stkp,d);
+#endif
 			}
 		}
 		if(arrmax)
@@ -2403,7 +2427,16 @@ static void mac_copy(register Mac_t *mp,register const char *str, register int s
 				if(n==S_MBYTE)
 				{
 					if(sh_strchr(mp->ifsp,cp-1)<0)
+					{
+						/*
+						 * The multi-byte character that was found has the same initial
+						 * byte as the IFS delimiter, but it's a different character. Put
+						 * the first byte onto the stack and continue; multi-byte characters
+						 * otherwise lose their initial byte.
+						 */
+						sfputc(stkp,c);
 						continue;
+					}
 					n = mbsize(cp-1) - 1;
 					if(n==-2)
 						n = 0;
diff --git a/src/cmd/ksh93/tests/variables.sh b/src/cmd/ksh93/tests/variables.sh
index c6e8c4203..6afe9022b 100755
--- a/src/cmd/ksh93/tests/variables.sh
+++ b/src/cmd/ksh93/tests/variables.sh
@@ -435,6 +435,39 @@ case $(unset IFS; set -- $v; print $#) in
 *)	err_exit 'BUG_KUNSETIFS detection failed'
 esac
 
+# Multi-byte characters should work with $IFS
+(
+	LC_ALL=C.UTF-8  # The multi-byte tests are pointless without UTF-8
+
+	# Test the following characters:
+	# Lowercase accented e  (two bytes)
+	# Roman sestertius sign (four bytes)
+	for delim in é 𐆘; do
+		IFS="$delim"
+		set : :
+		[ "$*" == ":$delim:" ] || err_exit "IFS failed with multi-byte character $delim (expected :$delim:, got $*)"
+
+		read -r first second third <<< "one${delim}two${delim}three"
+		[[ $first == one ]] || err_exit "IFS failed with multi-byte character $delim (expected one, got $first)"
+		[[ $second == two ]] || err_exit "IFS failed with multi-byte character $delim (expected two, got $second)"
+		[[ $third == three ]] || err_exit "IFS failed with multi-byte character $delim (expected three, got $three)"
+
+		# Ensure subshells don't get corrupted when IFS becomes a multi-byte character
+		expected_output="$(printf ":$delim:\\ntrap -- 'echo end' EXIT\\nend")"
+		output="$(LANG=C.UTF-8; IFS=$delim; set : :; echo "$*"; trap "echo end" EXIT; trap)"
+		[[ $output == $expected_output ]] || err_exit "IFS in subshell failed with multi-byte character $delim (expected $expected_output, got $output)"
+	done
+
+	# Multibyte characters with the same initial byte shouldn't be parsed as the same
+	# character if they are different. The regression test below tests two characters
+	# with the same initial byte (0xC2).
+	IFS='£'  # £ = C2 A3
+	v='abc§def ghi§jkl'  # § = C2 A7 (same initial byte)
+	set -- $v
+	v="${#},${1-},${2-},${3-}"
+	[[ $v == '1,abc§def ghi§jkl,,' ]] || err_exit "IFS treats £ (C2 A3) and § (C2 A7) as the same character"
+)
+
 # ^^^ end: IFS tests ^^^
 # restore default split:
 unset IFS