Mounting evidence indicates dryland ecosystems play an important role in driving the interannual variability and trend of the terrestrial carbon sink. Nevertheless, our understanding of the seasonal dynamics of dryland ecosystem carbon uptake through photosynthesis [gross primary productivity (GPP)] remains relatively limited due in part to the limited availability of long-term data and unique challenges associated with satellite remote sensing across dryland ecosystems. Here, we comprehensively evaluated longstanding and emerging satellite vegetation proxies in their ability to capture seasonal dryland GPP dynamics. Specifically, we evaluated: 1) reflectance-based proxies normalized difference vegetation index (NDVI), soil adjusted vegetation index (SAVI), near infrared reflectance index (NIR_v), and kernel NDVI (kNDVI) from the MODerate resolution Imaging Spectroradiometer (MODIS); and 2) newly available physiologically-based proxy solar-induced chlorophyll fluorescence (SIF) from the TROPOspheric Monitoring Instrument (TROPOMI). As a performance benchmark, we used GPP estimates from a robust network of 21 western United States eddy covariance tower sites that span representative gradients in dryland ecosystem climate and functional composition. We found that NIR_v and SIF were the best performing GPP proxies and captured complementary aspects of seasonal GPP dynamics across dryland ecosystem types. NIR_v offered better performance than the other proxies across relatively low-productivity, sparsely non-evergreen vegetated sites (R² = 0.59 ± 0.13); whereas SIF best captured seasonal dynamics across relatively high-productivity sites, including evergreen-dominated sites (R² = 0.74 ± 0.07). Notably, across grass-dominated sites, all reflectance-based proxies (NDVI, SAVI, NIR_v and kNDVI) showed significant seasonal bias (hysteresis) that strengthened with the total fraction of woody vegetation cover, likely due to seasonal patterns in woody vegetation reflectance that are unrelated to or decoupled from GPP. Future efforts to fully integrate the complementary strengths of NIR_v and SIF could significantly improve our understanding and representation of dryland GPP dynamics in satellite-based models.