The dynamical behavior of the Korean treasury bond (KTB) futures is investigated using a modified rescaled range (R/S) analysis. Lo's modified R/S analysis as well as classical Hurst's R/S statistics are utilized in order to analyze tick data of KTB futures. The Hurst exponent can be estimated by both classical and modified R/S statistics. The results of this study show that a series of returns has larger long-term correlation as the time lag increases. The Hurst exponent that shows a memory effect for the tick data of KTB futures is larger than 0.5, while the Hurst exponent has a value of nearly 0.5 for the corresponding shuffled data.