Objectives: To compare self-reported and observational work sampling techniques when applied to ward-based nurses. Methods: A self-reported work sampling study was undertaken with nine registered nurses in an Australian teaching hospital over 8.5 weeks, followed by an observational work sampling study conducted over 4.5 weeks. Both studies used a random reminder method and a multidimensional work task classifica-tion. Field notes were also recorded and analysed. Results: 3910 data points were collected, 667 during the self-report study and 3243 in the observational study. The two techniques yielded significant differences in work patterns of registered nurses. The observational study showed that compared with the self-reported study, patient care (40% versus 33%, P <0.000) and -ward-related activities (7% versus 3%; P <0.001) were recorded significantly more frequently, and documentation less frequently (8% versus 19%; P <0.000). Both the techniques generated similar proportions of time spent in breaks (12%), medication tasks (13%) and clinical discussion (15%). The self-report technique was poorly accepted by nursing staff. The observational technique was well accepted and data collection was more efficient. Conclusions: The self-report work sampling technique is not a reliable method for obtaining an accurate reflection of the work tasks of ward-based nurses. The observational technique was preferred by nurses, and despite concern regarding a potential Hawthorne effect, this was not substantiated.