Abstract: Based on direct economic losses and socioeconomic data, consumer price index (CPI) and conventional normalization method (CNM) and alternative normalization method (ANM) were used to investigate the damage caused by tropical cyclones which influenced the mainland of China in the last three decades, and the temporal and spatial variation of original and normalized economic losses were also analyzed. The results are as follows: (1) Original losses are helpful for comparison with other natural disasters occurring in the same year, and also suitable for updating datasets year by year. While normalized losses has an advantage of conducting a long-term study based on a homogeneous series, the CPI method is simple and convenient, and can be recommended to apply in China's disaster data processing and analysis. In regions with rapid growth of population and wealth, the CNM and ANM normalized losses might better address how the population and wealth exposures affect the loss, and thus may be more suitable for conducting a long-term study accordingly; (2) Totally 243 ITCs have influenced 22 provinces in the mainland of China during 1984 to 2014, and no significant monotonic trend was detected for the annual frequency, but an obvious upward trend is found for strong ITCs (with the intensity higher than typhoon); (3) The ITCs annual direct economic losses have shown significant increasing trends according to the original and CPI normalized series, while no apparent growing tendency is detected for the CNM or ANM series. The highest loss occurred in 2013 by the original series, but changed in 1996 according to the three normalized series, with an obvious abruption. However, a common feature of 2~3 year oscillation is found. Zhejiang, Guangdong and Fujian provinces rank the top three in all of the four loss series; (4) Herb (1996) is the severest tropical cyclone among the ten worst ITCs during the past 31 years, whereas the order of other single cyclones depends on the normalization method used.