Thermal spaceborne remote sensing has been used to study, monitor, and forecast volcanic activity for decades. But these data have not been used systematically at high spatial resolution to study changes in volcano temperatures across an entire region spanning multiple decades to understand background thermal activity and its relation to unrest and eruption. We have developed a first-of-a-kind database that uses manual analysis to identify and collect data for volcanic thermal output with 90 m/pixel spatial resolution for 330 potentially active volcanoes found in Latin America between the years 2000–2017. This database is reliant on the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) sensor due to its high spatial resolution data, capability of detecting low-level thermal features, its accessibility and reliability compared to similar data types, and the long time series available at multiple volcanoes. A total of 88 Latin American volcanoes were found to have some type of volcanic thermal feature detected by ASTER, and here we document thermal features at 16 of these volcanoes detected from space for the first time. We have recorded these thermal features, including the temperature above background, area above background, and the timing and location of detection in the ASTER Volcanic Thermal Output Database (AVTOD). By comprehensively analyzing such a large dataset, we are able to quantitatively analyze some of the issues with these data, including 24% of all volcanoes in this study failing to meet the acquisition rates proposed by the current acquisition plan and 44.5% of all acquisitions being unusable due to interference from clouds. We provide recommendations of how to update future acquisition plans that would focus on night-time and cloud-free acquisitions. In order to confirm the validity of AVTOD it was tested against the existing Moderate Resolution Imaging Spectroradiometer (MODIS)-based MIROVA database. In some cases, we found a high degree of correlation between the two datasets (r2 = 0.87). In other cases, however, correlation was limited due to the difference in spatial resolution of these two data types. By examining the maximum temperature detected by every volcano in the database we found 46 that never reach a temperature high enough to be detected by MODIS-class thermal sensors. The information in this database provides new insights about volcanic activity both on its own, and in combination with other data types, as well as a data-driven approach to improve key features in future sensors.