Accelerometers offer opportunities for researchers to capture validdata about the intensity and amount of physical behavior (PB) in real-time over a period of several days and weeks. From this multi-dimensional data, a great number of metrics can be derived to captureand describe the unique aspects of PB. The goal of this paper is to helpthe end-user of PB monitoring devices (novice to intermediate experi-ence) wade through sometimes excessive technical details of accel-erometry to outline best practices in selecting and applying devices toquantify three major behavioral categories of common interest to theresearch community: physical activity (PA), sedentary behavior (SB)and sleep. The effects of these decisions on the metrics (energy ex-penditure, activity intensity, body position, activity patterns) can occurin a variety of ways. The device, carrying position (hip, wrist, thigh)and recording parameters (epoch length (EL), frequency, memory ca-pacity, recording frequency andfilters) have a large influence on themeasured activity. The different backgrounds such as study design(purpose, repeated measurements) and duration (time frame, weartime) as well as data storage and evaluation must be taken into accountwhen determining the parameters. Finally, the evaluation must adjustseveral levers (raw data, context information, non-wear time, intensityclassification, compliance) depending on the target variables. Lookinginto the future, current developments in statistical analysis are dis-cussed, because the research community has not yet reached a con-sensus on the most promising approach. There are exciting develop-ments ahead of us in the future. Sleep in particular is increasingly beingseen as an influencing factor for health. Together with the technicaldevelopments in sensors which will become incrementally smaller,more accurate and in the near future will be integrated directly into ourclothes or skin, accelerometry is facing exciting times and lots of data toevaluate.