From 2016 to 2018, we surveyed the world’s largest natural history museum collections to begin mapping this globally distributed scientific infrastructure. The resulting dataset includes 73 institutions across the globe. It has: Basic institution data for the 73 contributing institutions, including estimated total collection sizes, geographic locations (to the city) and latitude/longitude, and Research Organization Registry (ROR) identifiers where available. Resourcing information, covering the numbers of research, collections and volunteer staff in each institution. Indicators of the presence and size of collections within each institution broken down into a grid of 19 collection disciplines and 16 geographic regions. Measures of the depth and breadth of individual researcher experience across the same disciplines and geographic regions. This dataset contains the data (raw and processed) collected for the survey, and specifications for the schema used to store the data. It includes: A diagram of the MySQL database schema. A SQL dump of the MySQL database schema, excluding the data. A SQL dump of the MySQL database schema with all data. This may be imported into an instance of MySQL Server to create a complete reconstruction of the database. Raw data from each database table in CSV format. A set of more human-readable views of the data in CSV format. These correspond to the database tables, but foreign keys are substituted for values from the linked tables to make the data easier to read and analyse. A text file containing the definitions of the size categories used in the collection_unit table. The global collections data may also be accessed at https://rebrand.ly/global-collections. This is a preliminary dashboard, constructed and published using Microsoft Power BI, that enables the exploration of the data through a set of visualisations and filters. The dashboard consists of three pages: Institutional profile: Enables the selection of a specific institution and provides summary information on the institution and its location, staffing, total collection size, collection breakdown and researcher expertise. Overall heatmap: Supports an interactive exploration of the global picture, including a heatmap of collection distribution across the discipline and geographic categories, and visualisations that demonstrate the relative breadth of collections across institutions and correlations between collection size and breadth. ...