Dynamic vegetation models have been used to assess the resilience of tropical forests to climate change, but the global application of these modeling experiments often misrepresent carbon dynamics at a regional level, limiting the validity of future projections. Here, a dynamic vegetation model (LPJ-GUESS) was adapted to simulate present day potential vegetation as a baseline for climate change impact assessments in the evergreen and deciduous forests of Bolivia. Results were compared to biomass measurements (819 plots), and remote sensing data. Using regional parameter values for allometric relations, specific leaf area, wood density and disturbance interval, a realistic transition from the evergreen Amazon to the deciduous dry forest was simulated. This transition coincided with threshold values for precipitation (1400 mm yr−1) and water deficit (i.e. potential evapotranspiration minus precipitation) (−830 mm yr−1), beyond which leaf abscission became a competitive advantage. Significant correlations were found between modeled and observed values of seasonal leaf abscission (R2 = 0.6, p <0.001) and vegetation carbon (R2 = 0.31, p <0.01). Modeled Gross Primary Productivity (GPP) and remotely sensed Normalized Difference Vegetation Index (NDVI) showed that dry forests were more sensitive to rainfall anomalies than wet forests. GPP was positively correlated to the El Niño Southern Oscillation index in the Amazon, and negatively correlated to consecutive dry days. Decreasing rainfall trends were simulated to reduce GPP in the Amazon. The current model set-up provides a baseline for assessing the potential impacts of climate change in the transition zone from wet to dry tropical forests in Bolivia.