Purpose: This paper carries out an empirical investigation of uncovered interest rate parity (UIP) model using both short- and long-horizon data for Australia and New Zealand. Key literature / theoretical perspective: The Uncovered Interest Rate Parity (UIP), one of the most popular approaches to assess the efficiency of the foreign exchange, has reported unfavourable results. Extensive surveys of the relevant literature by Froot and Thaler (1990 ), Taylor (1995 ), Lucio (2005 ), Chinn (2006 ), Isard (2006 ) reveal that the majority of studies, using a variety of estimation techniques, currencies and time periods, find the coefficient on the interest rate differential which is not only smaller than the theoretical value of unity but also displays the ‘wrong’ sign. Design/methodology/approach: In contrast to previous studies using OLS estimate that yields biased and inconsistent estimates in the present of an omitted risk premium, we apply GMM model that relates the risk premium to underlying economic variables. Findings: Our paper indicates that short-run horizon regression yield negative coefficients of about minus unity while three out of four coefficients yield positive values in the long-horizon regression. Research limitations/implications: Not all of the UIP hypothesises are tested. The paper may improve by carry out some more tests of these hypothesises and adjust the model in appropriate ways. Practical and Social implications: The thesis contribute to the current empirical study by focusing on the case of AU and NZ.