diff --git a/Number_types/include/CGAL/GMP/Gmpfr_type.h b/Number_types/include/CGAL/GMP/Gmpfr_type.h
index 1e4d362abbb..d16c76a8663 100644
--- a/Number_types/include/CGAL/GMP/Gmpfr_type.h
+++ b/Number_types/include/CGAL/GMP/Gmpfr_type.h
@@ -37,6 +37,7 @@
 #include <string>
 #include <limits>
 #include <CGAL/Uncertain.h>
+#include <CGAL/ipower.h>
 
 namespace CGAL{
 
@@ -861,28 +862,24 @@ std::pair<std::pair<double,double>,long> Gmpfr::to_interval_exp()const{
 
 inline
 std::pair<Gmpz,long> Gmpfr::to_integer_exp()const{
-        if(is_zero())
-                return std::make_pair(Gmpz(0),0);
-        Gmpz z;
-        long e=mpfr_get_z_exp(z.mpz(),fr());
-        CGAL_assertion_msg(mpfr_get_emin()<=e && mpfr_get_emax()>=e,
-                           "exponent out of range");
-        if(mpz_sgn(z.mpz())!=0&&mpz_tstbit(z.mpz(),0)==0){
-                unsigned long firstone=mpz_scan1(z.mpz(),0);
-                CGAL_assertion(mpz_divisible_2exp_p(z.mpz(),firstone)!=0);
-                Gmpz d(1);
-                mpz_mul_2exp(d.mpz(),d.mpz(),firstone);
-                CGAL_assertion(mpz_divisible_p(z.mpz(),d.mpz())!=0);
-                mpz_divexact(z.mpz(),z.mpz(),d.mpz());
-                e+=firstone;
-                CGAL_assertion(mpfr_get_emax()>=e);
-        }
-        CGAL_assertion_code(Gmpfr::Precision_type p=mpz_sizeinbase(z.mpz(),2);)
-        CGAL_assertion_code(if(MPFR_PREC_MIN>p) p=MPFR_PREC_MIN;)
-        CGAL_assertion_code(Gmpfr test(z,p);)
-        CGAL_assertion_code(mpfr_mul_2si(test.fr(),test.fr(),e,GMP_RNDN);)
-        CGAL_assertion_msg(mpfr_equal_p(test.fr(),fr())!=0,"conversion error");
-        return std::make_pair(z,e);
+
+  if(this->is_zero()) 
+    return std::make_pair(Gmpz(0),long(0));
+  
+  Gmpz z;
+  long e=mpfr_get_z_exp(z.mpz(),this->fr());
+  
+  long zeros = mpz_scan1(z.mpz(),0);
+  z >>= zeros;
+  e +=  zeros;
+  
+  CGAL_postcondition(z % 2 != 0);
+  CGAL_postcondition_code(if (e >= 0))
+    CGAL_postcondition( (*this) == (Gmpfr(z)) * CGAL::ipower(Gmpfr(2), e));
+  CGAL_postcondition_code(else)
+    CGAL_postcondition( (*this) == (Gmpfr(z)) / CGAL::ipower(Gmpfr(2),-e));
+  
+  return std::make_pair(z,e);
 }
 
 inline