Lessons learned from data and information handling for the widely used Comprehensive Ocean-Atmosphere Data Set (COADS) will be discussed. In COADS, marine meteorological observations from voluntary observing ships date back (in the latest data release) to 1792. These are combined with surface-level oceanographic data since the late 19th century, and with data from drifting and moored buoys and other automated platform types for recent decades. In the first stage of COADS processing, this diverse collection of data is converted into a common format. Then quality control (QC) is applied, and duplicate marine reports (e.g., received from real-time and delayed-mode sources) are eliminated. Finally, higher-level products are created in the form of global monthly summary statistics for 2-degree latitude x 2-degree longitude boxes (and for 1-degree x 1-degree boxes since 1960). The QC’d observations, higher-level products, and supporting metadata are readily available, and receive unrestricted distribution to the international user community. The approaches to these challenging data handling tasks have evolved with technology since the inception of the project in 1981, and more improvements are needed. We discuss our progress on the improvements to the climatological QC screening and other changes that will continue to help to assure data quality and continuity.