Overcrowding is a phenomenon commonly observed at emergency departments (EDs) in many hospitals, and negatively impacts patients, healthcare professionals and organisations. Health care organisations are expected to act proactively to cope with a high patient volume by understanding and predicting the patterns of ED presentations. The aim of this study was, therefore, to identify the patterns of patient flow at a paediatric ED in order to assist the management of EDs. Data for ED presentations were collected from the Royal Children's Hospital in Melbourne, Australia, with the time-frame of July 2003 to June 2008. A linear regression analysis with trigonometric functions was used to identify the pattern of patient flow at the ED. The results showed that a logarithm of the daily average ED presentations was increasing exponentially (as explained by 0.004t + 0.00005t2 with t representing time, p<0.001). The model also indicated that there was a yearly oscillation in the frequency of ED presentations, in which lower frequencies were observed in summer and higher frequencies during winter (as explained by -0.046 sin(2(pi)t/12)-0.083 cos(2(pi)t/12), p<0.001). In addition, the variation of the oscillations was increasing over time (as explained by -0.002t*sin(2(pi)t/12)-0.001t*cos(2(pi)t/12), p<0.05). The identified regression model explained a total of 96% of the variance in the pattern of ED presentations. This model can be used to understand the trend of the current patient flow as well as to predict the future flow at the ED. Such an understanding will assist health care managers to prepare resources and environment more effectively to cope with overcrowding.