diff --git a/Packages/Interval_arithmetic/include/CGAL/Interval_arithmetic.h b/Packages/Interval_arithmetic/include/CGAL/Interval_arithmetic.h
index 3304b20fe49..6d257b1d23f 100644
--- a/Packages/Interval_arithmetic/include/CGAL/Interval_arithmetic.h
+++ b/Packages/Interval_arithmetic/include/CGAL/Interval_arithmetic.h
@@ -535,7 +535,12 @@ CGAL_number_type_tag(CGAL_Interval_nt_advanced)
 // Finally we source the CGAL_convert_to<CGAL_Interval_nt_advanced>()
 // functions from other NTs, when necessary.
 // CGAL_convert_to<CGAL_Interval_nt>() is templated below, only specialized
-// for double.
+// for double (handled genericaly anyway).
+
+// The following big template should be put in number_utils.h, imho.
+template <class NTtarget, class NTsource>
+inline NTtarget CGAL_convert_to (const NTsource & source)
+{ return NTtarget(source); }
 
 #ifdef CGAL_GMPZ_H
 #include <CGAL/Interval_arithmetic/IA_Gmpz.h>
@@ -562,7 +567,7 @@ CGAL_number_type_tag(CGAL_Interval_nt_advanced)
 #endif
 
 template <class FT>
-inline CGAL_Interval_nt CGAL_convert_to<CGAL_Interval_nt>(const FT &z)
+inline CGAL_Interval_nt CGAL_convert_to(const FT &z)
 {
     CGAL_FPU_set_rounding_to_infinity();
     CGAL_Interval_nt tmp(CGAL_convert_to<CGAL_Interval_nt_advanced>(z));
@@ -576,12 +581,10 @@ inline CGAL_Interval_nt CGAL_convert_to<CGAL_Interval_nt>(const FT &z)
 // inline ET CGAL_convert_to (const CT & ct)
 // { return ET(ct); }
 
-inline CGAL_Interval_nt_advanced CGAL_convert_to<CGAL_Interval_nt_advanced>
-	(const double d)
-{ return CGAL_Interval_nt_advanced(d); }
+// inline CGAL_Interval_nt_advanced CGAL_convert_to (const double d)
+// { return CGAL_Interval_nt_advanced(d); }
 
-inline CGAL_Interval_nt CGAL_convert_to<CGAL_Interval_nt>
-	(const double d)
-{ return CGAL_Interval_nt(d); }
+// inline CGAL_Interval_nt CGAL_convert_to (const double d)
+// { return CGAL_Interval_nt(d); }
 
 #endif // CGAL_INTERVAL_ARITHMETIC_H