diff --git a/src/Makefile b/src/Makefile
index 911e533..e7db948 100644
--- a/src/Makefile
+++ b/src/Makefile
@@ -18,7 +18,7 @@ unittests: unittest_mime.o unittest_nfc.o unittest_timestamp.o \
     unittest_stringcase.o unittest_toutf8.o unittest_address.o \
     unittest_rule.o unittest_subject.o \
     gtest-all.o gtest_main.o libpEpMIME.a
-	${CXX} -L${HOME}/local/lib/ -o $@ $^ -lpEpAdapter -lpEpEngine -lpthread
+	${CXX} -L${HOME}/local/lib/ -o $@ $^ -lpEpAdapter -lpEpEngine -lpthread -liconv
 
 gtest-all.o: $(GTEST_DIR)/src/gtest-all.cc
 	$(CXX) $(CPPFLAGS) -I$(GTEST_DIR) $(CXXFLAGS) -isystem $(GTEST_DIR)/include -o $@ -c $<
diff --git a/src/to_utf8.cc b/src/to_utf8.cc
index f7de5c0..f179d30 100644
--- a/src/to_utf8.cc
+++ b/src/to_utf8.cc
@@ -3,6 +3,8 @@
 #include "nfc.hh"  // for toUtf8() of single codepoints. :-)
 #include <algorithm>
 #include <boost/algorithm/string/case_conv.hpp>
+#include <iconv.h>
+#include <memory>
 
 namespace
 {
@@ -58,23 +60,73 @@ namespace
         }
         return ret;
     }
-}
 
-std::string to_utf8( const std::string& charset, const std::string& s)
+
+static const size_t IconvBufSize = 64;
+
+std::string to_utf8_iconv(const std::string& charset, const std::string& s)
 {
-    switch( case_hash(charset) )
+    iconv_t ict = iconv_open("UTF-8", charset.c_str());
+    if(ict == (iconv_t)-1)
     {
-        case "UTF-8"_case :
-        case "UTF8"_case  : return s;
-        case "ISO-8859-1"_case: return from_latin1(s);
+        if(errno==EINVAL)
+        {
+            throw std::runtime_error("Cannot convert from charset \"" + charset + "\" to UTF-8.");
+        }else{
+            throw std::runtime_error(std::string("Internal error: ") + strerror(errno) );
+        }
     }
-    // try again with to_upper;
+    
+    // be exception-safe from here on:
+    auto ict_wrapper = std::unique_ptr<void, decltype(&iconv_close)>( ict, &iconv_close);
+    
+    std::string ret;
+    ret.reserve(s.size());
+    
+    char buffer[ IconvBufSize ];
+    char* in_p = const_cast<char*>(s.c_str());  // iconv sucks.
+    size_t in_len = s.size();
+    
+    while(in_len)
+    {
+        char* out_p = buffer;
+        size_t out_len = IconvBufSize;
+        const size_t r = iconv(ict, &in_p, &in_len, &out_p, &out_len);
+        if(r==static_cast<size_t>(-1))
+        {
+            if(errno == E2BIG)
+            {
+                // ignore
+            }else{
+                // skip octet
+                ++in_p;
+                --in_len;
+            }
+        }
+        ret.append(buffer, buffer + IconvBufSize - out_len);
+    }
+    
+    return ret;
+}
+
+
+} // end of anonymous namespace
+
+
+std::string to_utf8( const std::string& charset, const std::string& s)
+{
     std::string charset_upper{charset};
     boost::algorithm::to_upper(charset_upper);
-    if(charset_upper == charset)
+
+    switch( lcase_hash(charset_upper) )
     {
-        throw std::runtime_error("Unknown charset \"" + charset + "\"");
+        case "UTF-8"_lcase :
+        case "UTF8"_lcase  : return s;
+        case "CP1252"_lcase:
+        case "CP_1252"_lcase:
+        case "ISO-8859-1"_lcase: return from_latin1(s);
     }
-    return to_utf8( charset_upper, s);
+    // all other charsets: let's do that by libiconv. :-/
+    return to_utf8_iconv(charset, s);
 }
 
diff --git a/src/unittest_toutf8.cc b/src/unittest_toutf8.cc
index de263d7..ac50264 100644
--- a/src/unittest_toutf8.cc
+++ b/src/unittest_toutf8.cc
@@ -54,3 +54,15 @@ TEST( ToUtf8, Latin1 )
 	EXPECT_EQ( to_utf8("ISO-8859-1", "\x84\xdc" "bergr\xf6\xdf" "en\xe4" "nderung\x93: 10\x80!"), "„Übergrößenänderung“: 10€!" );
 	EXPECT_EQ( to_utf8("UTF-8", "„Übergrößenänderung“: 10€!"), "„Übergrößenänderung“: 10€!" );
 }
+
+TEST( ToUtf8, Latin9 )
+{
+	EXPECT_EQ( to_utf8("ISO-8859-15", "\xdc" "bergr\xf6\xdf" "en\xe4nderung: 10\xa4!"), "Übergrößenänderung: 10\xe2\x82\xac!" );
+	std::string latin9, utf8;
+	for(unsigned u=0; u<277; ++u)
+	{
+		latin9 += "\xb0\xa4\xbe"; // degree, euro, capital Y with diaeresis: °€Ÿ
+		utf8 += "\xc2\xb0" "\xe2\x82\xac" "\xc5\xb8";
+		EXPECT_EQ( to_utf8("ISO-8859-15", latin9), utf8 );
+	}
+}