This study looks at how academic libraries, especially research oriented institutions, develop and use cost per download data in collection decision-making. The study is based on data from 52 institutions, predominantly from the USA but also from Canada, the UK , continental Europe and elsewhere. Data in the report is broken out by type of institution (i.e. research university, doctoral-level, etc.) and by overall student enrollment, tuition, for public and private institutions and for those located in the USA and all other countries. Data is also presented separately for collections oriented towards healthcare and medicine, and for multidisciplinary collections.The 54-page study helps its readers to answer questions such as: How precise an idea do libraries have about the cost per download of their subscribed journals? How many libraries feel that they measure this cost well? What tools, applications or programs do they use to obtain or develop this data? What makes it easier or harder to obtain such data? How much confidence do they have in the accuracy of the data often made available by journals publishers? Do some of these publishers produce more reliable data than others? If so , which ones? Does the library use benchmarking data from other libraries or consortia when developing or using their in-hour cost per download data? Exactly what is the cost per download for the library's most and least expensive journals subscription packages? Is the library making any special efforts to obtain or obtain better cost per download data as a result of the COVID-19 pandemic and ensuing pressure on library budgets?