Background Time nurses spend with patients is associated with improved patient outcomes, reduced errors, and patient and nurse satisfaction. Few studies have measured how nurses distribute their time across tasks. We aimed to quantify how nurses distribute their time across tasks, with patients, in individual tasks, and engagement with other health care providers; and how work patterns changed over a two year period. Methods Prospective observational study of 57 nurses for 191.3 hours (109.8 hours in 2005/2006 and 81.5 in 2008), on two wards in a teaching hospital in Australia. The validated Work Observation Method by Activity Timing (WOMBAT) method was applied. Proportions of time in 10 categories of work, average time per task, time with patients and others, information tools used, and rates of interruptions and multi-tasking were calculated. Results Nurses spent 37.0%[95%CI: 34.5, 39.3] of their time with patients, which did not change in year 3 [35.7%; 95%CI: 33.3, 38.0]. Direct care, indirect care, medication tasks and professional communication together consumed 76.4% of nurses' time in year 1 and 81.0% in year 3. Time on direct and indirect care increased significantly (respectively 20.4% to 24.8%, P < 0.01;13.0% to 16.1%, P < 0.01). Proportion of time on medication tasks (19.0%) did not change. Time in professional communication declined (24.0% to 19.2%, P < 0.05). Nurses completed an average of 72.3 tasks per hour, with a mean task length of 55 seconds. Interruptions arose at an average rate of two per hour, but medication tasks incurred 27% of all interruptions. In 25% of medication tasks nurses multi-tasked. Between years 1 and 3 nurses spent more time alone, from 27.5%[95%CI 24.5, 30.6] to 39.4%[34.9, 43.9]. Time with health professionals other than nurses was low and did not change. Conclusions Nurses spent around 37% of their time with patients which did not change. Work patterns were increasingly fragmented with rapid changes between tasks of short length. Interruptions were modest but their substantial over-representation among medication tasks raises potential safety concerns. There was no evidence of an increase in team-based, multi-disciplinary care. Over time nurses spent significantly less time talking with colleagues and more time alone.