With the liberalization in capital flows and the advances in telecommunication and computer technology, international portfolio selection has been becoming a hot topic for both practitioners and researchers. In traditional international portfolio selection, security selection used to be decided first without considering the fluctuation of foreign exchange rates. Then hedging strategies were made to eliminate the fluctuation of foreign exchange rates. In addition, the security returns and the foreign exchange rates were usually assumed to be random variables. However, in reality, it is usually difficult for many currencies except major currencies to find suitable instruments to make effective currency hedging though cross-hedging may be of some help. In addition, hedging strategy may not be a good choice because the foreign exchange rate fluctuation may also bring the high return to the investors. Furthermore, there are many researches showing that security returns may not be random sometimes. In this paper, we discuss international portfolio selection problem with both foreign exchange rates and security returns containing a new type of uncertainty which is neither random nor fuzzy. Based on the latest development on uncertainty theory, we develop new mean-variance models considering the uncertain foreign exchange rates and security returns simultaneously. To solve the new models in general cases, a hybrid intelligent algorithm is provided. As an illustration, an example is also presented.