Biomass fuelled integrated gasification/gas turbines (BIG/GTs) have been found to be one of the most promising technologies to maximise electricity output in the sugar industry. However, biomass fuels contain alkali metals (Na and K) which may be released during the gasification processes and cause deleterious effects on the downstream hardware (e.g. the blades of gas turbines). Much research has therefore been focused on different kinds of gas cleaning. Most of these projects are using a fluidised bed gasifier and includes extensive gas cleaning which leads to a high capital investment. Increasing alkali retention/separation during the gasification may lead to improved producer gas quality and reduced costs for gas cleaning. However, very little quantitative information is available about the actual potential of this effect. In the present work, comparative bench-scale tests of bagasse gasification were therefore run in an isothermal fluidised bed gasifier and in a cyclone gasifier to evaluate which gasification process is most attractive as regards alkali retention/separation, and to try to elucidate the mechanisms responsible for the retention. The alkali retention in the fluidised bed gasifier was found to be in the range of 12-4% whereas in the cyclone gasifier the alkali separation was found to be about 70%. No significant coating of the fluidised bed's bed material particles could be observed. The SEM/EDS and the elemental maps of the bed material show that a non-sticky ash matrix consisting of mainly Si, Al and K were distributed in a solid form separated from the particles of bed material. This indicates the formation of a high temperature melting potassium containing silicate phase, which is continuously scavenged and lost from the bed through elutriation.