diff --git a/common/CMakeLists.txt b/common/CMakeLists.txt
index 91d79493..a4008fd1 100644
--- a/common/CMakeLists.txt
+++ b/common/CMakeLists.txt
@@ -7,6 +7,7 @@ list(APPEND _LIBUI_SOURCES
 	common/matrix.c
 	common/shouldquit.c
 	common/userbugs.c
+	common/utf.c
 )
 set(_LIBUI_SOURCES ${_LIBUI_SOURCES} PARENT_SCOPE)
 
diff --git a/common/uipriv.h b/common/uipriv.h
index d6b54e89..f22a08a6 100644
--- a/common/uipriv.h
+++ b/common/uipriv.h
@@ -1,4 +1,7 @@
 // 6 april 2015
+// TODO can extern "C"s nest?
+#include "utf.h"
+
 #ifdef __cplusplus
 extern "C" {
 #endif
diff --git a/windows/utf16.cpp b/windows/utf16.cpp
index 98954d0a..6271fff7 100644
--- a/windows/utf16.cpp
+++ b/windows/utf16.cpp
@@ -3,48 +3,42 @@
 
 // see http://stackoverflow.com/a/29556509/3408572
 
-#define MBTWC(str, wstr, bufsiz) MultiByteToWideChar(CP_UTF8, 0, str, -1, wstr, bufsiz)
-
 WCHAR *toUTF16(const char *str)
 {
 	WCHAR *wstr;
-	int n;
+	WCHAR *wp;
+	size_t n;
+	uint32_t rune;
 
 	if (*str == '\0')			// empty string
 		return emptyUTF16();
-	n = MBTWC(str, NULL, 0);
-	if (n == 0) {
-		logLastError(L"error figuring out number of characters to convert to");
-		return emptyUTF16();
-	}
-	wstr = (WCHAR *) uiAlloc(n * sizeof (WCHAR), "WCHAR[]");
-	if (MBTWC(str, wstr, n) != n) {
-		logLastError(L"error converting from UTF-8 to UTF-16");
-		// and return an empty string
-		*wstr = L'\0';
+	n = utf8UTF16Count(str, 0);
+	wstr = (WCHAR *) uiAlloc((n + 1) * sizeof (WCHAR), "WCHAR[]");
+	wp = wstr;
+	while (*str) {
+		str = utf8DecodeRune(str, 0, &rune);
+		n = utf16EncodeRune(rune, wp);
+		wp += n;
 	}
 	return wstr;
 }
 
-#define WCTMB(wstr, str, bufsiz) WideCharToMultiByte(CP_UTF8, 0, wstr, -1, str, bufsiz, NULL, NULL)
-
 char *toUTF8(const WCHAR *wstr)
 {
 	char *str;
-	int n;
+	char *sp;
+	size_t n;
+	uint32_t rune;
 
 	if (*wstr == L'\0')		// empty string
 		return emptyUTF8();
-	n = WCTMB(wstr, NULL, 0);
-	if (n == 0) {
-		logLastError(L"error figuring out number of characters to convert to");
-		return emptyUTF8();
-	}
-	str = (char *) uiAlloc(n * sizeof (char), "char[]");
-	if (WCTMB(wstr, str, n) != n) {
-		logLastError(L"error converting from UTF-16 to UTF-8");
-		// and return an empty string
-		*str = '\0';
+	n = utf16RuneCount(wstr, 0);
+	str = (char *) uiAlloc((n + 1) * sizeof (char), "char[]");
+	sp = str;
+	while (*wstr) {
+		wstr = utf16DecodeRune(wstr, &rune);
+		n = utf8EncodeRune(rune, sp);
+		sp += n;
 	}
 	return str;
 }
@@ -92,6 +86,8 @@ WCHAR *vstrf(const WCHAR *format, va_list ap)
 	return buf;
 }
 
+// TODO merge the following two with the toUTF*()s?
+
 // Let's shove these utility routines here too.
 // Prerequisite: lfonly is UTF-8.
 char *LFtoCRLF(const char *lfonly)