diff --git a/src/CLR/CorLib/corlib_native_System_Convert.cpp b/src/CLR/CorLib/corlib_native_System_Convert.cpp
index 8b6d693e47..a925e9e96b 100644
--- a/src/CLR/CorLib/corlib_native_System_Convert.cpp
+++ b/src/CLR/CorLib/corlib_native_System_Convert.cpp
@@ -25,9 +25,29 @@ HRESULT Library_corlib_native_System_Convert::NativeToInt64___STATIC__I8__STRING
         long long minValue = stack.Arg2().NumericByRef().s8;
         long long maxValue = stack.Arg3().NumericByRef().s8;
 
+        // normally we check if the result is in the given range
+        bool check = true;
+        
+        // Int64? => use also strtoull the result will be casted to Int64
+        if (isSigned && maxValue == 0x7FFFFFFFFFFFFFFF) isSigned = false;
+
+        // UInt64? => use also strtoull the result will be casted to Int64 and bypass the range check
+        if (minValue == 0 && maxValue == 0)
+        {
+            isSigned = false;
+            check = false;
+        }
+
+        // convert via strtoll / strtoull
         result = isSigned ? strtoll(str, nullptr, radix) : (long long) strtoull(str, nullptr, radix);
 
-        stack.SetResult_I8 ((result > maxValue || result < minValue) ? zero : result);
+        // the signed values for SByte, Int16 and Int32 are always positive for base 2, 8 or 16 conversions
+        // because the 64-bit function strtoll is used; need the post process the value
+        // if the result is greater max and smaller (max + 1) * 2 this value should be subtracted
+        if (isSigned && result > maxValue && result < (maxValue + 1) * 2) result -= (maxValue + 1) * 2;
+        
+        // for UInt64 the check will be bypassed
+        stack.SetResult_I8 ((check && (result > maxValue || result < minValue)) ? zero : result);
         
       #else
         // support for conversion from base 10 and 16 (partial)