Observational methods are under-utilised in nursing research, but have a number of benefits. This paper outlines two approaches to collecting observational data, namely continuous sampling and time sampling. Time sampling is critically reviewed, and methodological considerations when adopting this method are discussed. A major issue is the reliability in estimating frequency and duration of observed behaviours. Researchers have shown that the selection of the appropriate interval length is critical to this reliability, emphasising the need for pilot work to define and estimate average duration of behaviours to be observed. The paper outlines conditions when either of the two sampling methods, partial interval recording and momentary time sampling, may be reasonably selected. However, continuous sampling may be preferable for observing complex behaviours involving at least two participants.