All data generated and processed in scientific activities are considered research data. They constitute the basis of the research results. Data types can be quite diverse and may comprise measurements, secondary analyses, visualisations, models, and the results of polls and surveys. Equally manifold are the file formats used to hold numbers, text, code, or graphics.  Even physical samples such as minerals or tissues are sometimes defined as "data".

It is good scientific practice to keep at least those data underlying published works for no less than ten years. If as many data as possible are also made publically accessible, they can be re-used in further projects. Also scientific results are easier to validate. This is why funders and important international journals increasingly demand that research data are not only kept but also published.

The extent to which costs for research data management measures arise is different in each project. Two example calculations can be found here. Important influencing factors are the volume of data, the number of files and the degree of homogeneity. The more manual work is required, the higher the labor requirement and thus the costs. The publication of data is free of charge in many repositories. However, larger volumes of data or the use of additional services (such as data preparation and curation) may incur costs.