Double byte characters in the URL - best avoided?

[Red_Mud_Rookie]

We are doing some optimisation on sites in the APAC region, namely China, Hong Kong, Taiwan and Japan.

We have set the url generator to automatically use the heading of the page in the URL which works fine for countries using Latin characters, but is causing problems, particularly in IE, when it comes to the double byte countries.

For some reason, IE struggles with double byte and displays URLs in their rather ugly, coded form.

Anybody got any suggestions on whether we should persist with the keyword URLs or revert to the non-descriptive URLs for the double byte countries?

The reason I ask is it's a balance of SEO benefit vs not scaring IE users off with ugly URLs that look dreadful and spammy.

[ninjamarketer]

Search engines are placing more importance to usability than ever before and longer URLs are usually not considered user friendly. The popularity of URL shortening service is another proof that having shorter url is better.

The idea is to maintain the right balance between keyword usage in the URL and the length of URL.

[Red_Mud_Rookie]

As far as the search engines are concerned, the URLs are nice and clean with relevant keywords, but if a user browses the site in IE then the URLs do not maintain their double byte structure and instead convert to ugly URLs like this: mydomain.com/hk/hotels/asia-pacific/-中國-重慶-579-重慶-Sofitel-2-1-晚優惠-連早餐-1025557/