In this paper, we discuss how an existing empirical drought index, i.e. the Keetch-Byram Drought Index (KBDI) that is commonly used for assessing forest fire danger, has been adjusted and modified for improved use in tropical wetland ecosystems. The improvement included: (i) adjustment of the drought factor to the local climate, and (ii) addition of the water table depth as a dynamic factor to control the drought index. We distinguished three different indices, the original Keetch-Byram Drought Index, the adjusted KBDI (KBDIadj) that represents the original drought index, but including local climate information, and the modified KBDI (mKBDI) that considers both local climate information, and soil and hydrological characteristics. The mKBDI was developed and tested in a wetland forest of South Sumatra (Indonesia) from April 2009 to March 2011. During this period, hydrometeorological data were monitored and used to calculate the KBDI, KBDIadj, and mKBDI. First, mKBDI was calibrated using observed soil moisture that was converted to an observed drought index (DIobs). The results indicate that performance of the mKBDI is encouraging based on the following: (i) its pattern followed the dynamics of DIobs, (ii) prediction of frequency of fire danger classes, and (iii) statistically criteria. The mKBDI clearly outperformed KBDI and KBDIadj. Furthermore, we found a critical water table depth when it reaches maximum fire danger (0.85m for the wetland forest of South Sumatra) below which danger does not increase anymore. The mKBDI could be more widely applied, if pedotransfer functions are developed that link easily obtainable soil properties to the parameters of the water table factor. Our findings encourage land use planners, water managers and stakeholders (e.g. forest estate owners) to integrate local climate information, and soil and hydrological characteristics into the Keetch-Byram Drought Index to better predict fire danger, particularly in tropical wetland ecosystems.