This paper examines the empirical validity of the hypothesis of uncovered interest parity (UIP) using data from five Central and Eastern European countries with floating exchange rates for the period 2003 to 2014. The analysis includes forward‐looking as well as static expectations and allows for different types of structural break. The variable depicting the deviation from strict UIP is stationary when expectations are forward looking, suggesting that it is not possible to reject the UIP hypothesis with a constant risk premium. The deviation from strict UIP is however typically not stationary when expectations are static, even when structural breaks are incorporated, leading to the rejection of the UIP hypothesis with a constant risk premium. The results underscore the central role of expectations for the UIP hypothesis.