This dataset consists of over 800 field observations of ground failure (landslides, lateral spreading, and liquefaction) and other damage triggered by the 2019-2020 Puerto Rico earthquake sequence. The sequence started with a M4.7 earthquake on 28 December 2019, followed by many more earthquakes, including 15 larger than M5 (as of 7 July 2020). The M6.4 mainshock, which is thought to have triggered much of the observed ground failure, occurred on 7 January 2020. Most field reconnaissance efforts documented here took place as soon as possible after the mainshock, from 12-18 January 2020, to attempt to capture ephemeral data before evidence was destroyed by natural forces or repairs, but observations continued to be made through the end of February 2020. To organize the data, we have assigned each of the ~800 observations to a single ground failure type category that best describes the observation. The observations are symbolized by this main category in the accompanying ArcGIS Online Dashboard to make it easy to visualize and browse the observations. Main categories include lateral spreading, liquefaction, disrupted slides and falls, soil landslides, ground cracks, subsidence, damage, null (no ground failure) and other. We provide specific descriptions of these categories subsequently. The geotagged photos associated with most observation points include additional tags to more completely describe the observation. Many observations are also accompanied by text descriptions made during the field work by the observer. The ___location of the majority of the points represents the ___location of the observer/photographer because the coordinates come from geotagged photos and waypoints, but in a few cases, the locations of a distant occurrence of ground failure were estimated by the observer; these cases are noted in the comments. For inventories of landslides and liquefaction where the features are in the ___location of the occurrence, refer to the landslide inventory and liquefaction inventory that were derived in part from these field observations. The data is available in several file formats. The geodatabase (PR_observations_gdb.zip) includes the point locations along with a relationship class for relating attachments to the points. In the PR2020_observation_data.zip zip file, we also include a stand-alone shapefile (Shapefile/PR_observation_pts.shp) of the point locations. This can be used with the folder of photos (Photos) along with the photo table (Table/photo_table.csv) to relate photos to the shapefile. We were not able to investigate the entirety of the potentially affected area due to access and time restrictions as well as dense vegetation. Instead, the areas we investigated were guided by 1) the USGS ground failure models for the mainshock, 2) early satellite mappings of ground failures by colleagues, 3) reports of ground failure in the news, social media, and from emergency managers passed on to us by colleagues, and 4) reports from local residents and emergency managers in the field. In addition to documenting where ground failure did occur, another primary objective of our work was to document where ground failure did not occur even though susceptible conditions were present (null observations). An observation is categorized as a “Lateral spread” if there was displacement on gentle slopes, (commonly towards a water body and a result of liquefaction) regardless of whether there was sand ejecta or other direct evidence of liquefaction. Liquefaction is reserved for sand boils, fissures with ejecta, and other features that indicate liquefaction occurred but without obvious horizontal displacements towards bodies of water. Ground cracks without any obvious ties to liquefaction or lateral spreading are simply categorized as “Cracks.” Cracks are extensional except when otherwise noted in the “General notes” field. Ambiguous observations, such as crab burrows that were difficult to differentiate from sand boils, are categorized under “Other.” Water main damage and uplifted utility covers and pipes are generally categorized as “Damage” even though they are commonly related to liquefaction or lateral spreading. We only labeled such cases as liquefaction if direct evidence was present. Though documenting structural damage was not the primary objective of these field campaigns, we did collect some observations of building damage and these are included and categorized as “Damage”, but they are not comprehensive. “Subsidence” observations are based on observations of flooding and are likely due to tectonic deformation. We use the general term “disrupted slides and falls,” following Keefer (1984) for all landslides with high degrees of internal disruption. The majority of disrupted landslides were rock falls, but some were rock slides or involved debris and soil. We only noted one area that is classified as “Coherent slides” where there was a cut and fill slope failure.