Vegetation optical depth (VOD) from satellite passive microwave sensors has enabled monitoring of aboveground biomass carbon dynamics by building a relationship with static carbon maps over space and then applying this relationship to VOD time series. However, uncertainty in this relationship arises from changes in water stress, as VOD is mainly determined by vegetation water content, which varies at diurnal to interannual scales, and depends on changes in both biomass and relative moisture content. Here, we studied the reliability of using VOD from various microwave frequencies and temporal aggregation methods for estimating decadal biomass carbon dynamics at the global scale. We used the VOD diurnal variations to represent the magnitude of vegetation water content buffering caused by climatic variations for a constant amount of dry biomass carbon. This magnitude of VOD diurnal variations was then used to evaluate the likelihood of VOD decadal variations in reflecting decadal dry biomass carbon changes. We found that SMOS-IC L-VOD and LPDR X-VOD can be reliably used to estimate decadal carbon dynamics for 76.7% and 69.9% of the global vegetated land surface, respectively, yet cautious use is warranted for some areas such as the eastern Amazon rainforest. Moreover, the annual VOD aggregated from the 95% percentile of the nighttime VOD retrievals was proved to be the most suitable parameter for estimating decadal biomass carbon dynamics among the temporal aggregation methods. Finally, we validated the use of annual VOD for estimating interannual carbon dynamics by comparing VOD changes between adjacent years against eddy covariance estimations of gross primary production from flux sites over several land cover classes across the globe. Despite the large difference in spatial scales between them, the positive correlation obtained supports the capability of satellite VOD in quantifying interannual carbon dynamics.