Given the effort and time demanded for manually mapping forest cover change, tools that can automate this process are essential for monitoring large areas. The CLASlite algorithm is one of many tools that can be used for automated forest mapping, utilizing fractional cover to generate a subpixel analysis. In areas such as the várzea floodplains of the Amazon basin, the high discharges and sediment loads, together with the shallow slopes, generate a highly dynamic landscape due to fluvial processes. Our study therefore evaluated the accuracy of the CLASlite forest mapping algorithm in determining forest cover changes caused by fluvial dynamics, by comparing automated results with manually generated maps, both based on a time series of five Landsat 5/TM images, from 2005 to 2009. The study site was a small subset of the Mamirauá Sustainable Development Reserve, located in the Solimões/Amazon river floodplain, Brazil. By testing different combinations of algorithm parameters for the CLASlite forest cover mapping tool, we were able to determine how each parameter influenced mapping accuracy. Overall, the CLASlite algorithm has been shown to be very effective in mapping already consolidated forest areas, but efficiency decreases for areas closer to the riverbed, which are the most susceptible to river dynamics.