Determining research-related costs from intervention-related costs is important for both clinical and health services research. Often this task involves estimating what proportion of the workday personnel spend on a variety of activities. We used a portable random reminder beeper to measure the daily work activities and contacts of study nurses within the context of a multi-site randomized trial designed to assess the effectiveness of primary care. Nurses recorded 4920 work activities over 140 consecutive workdays. Research-related activities consumed the largest proportion of the workday, 42.5% (95% CI, 38.1-46.7) followed by patient care, 28.8% (24.1-33.2), personal time 16.4% (12.0-20.7), and time spent in transit 12.5% (9.1-15.9). Because this research-related time is spent performing tasks specific to the enrollment of patients and measurement of outcome variables, we will use an adjusted annual salary for these nurses (from $56,392-$32,425) when attributing costs of the intervention in cost-effectiveness analyses and for future management projections. Work sampling is a flexible, inexpensive method that was well accepted by the nurses in this study. Our results provide important insights into the costs analysis of complex interventions involving health professionals and may allow us to explore why the intervention worked or did not work at individual sites.