We formulated the hypothesis that lunar phases, identified by the fraction of the illuminated visible surface of the moon, have a relationship with the frequency of victims of aggression seen in an emergency department. If such a relationship exists, an increase in the frequency of incidents with the phases of full moon or new moon would be expected. In order to test this hypothesis, the daily frequency of victims of violent behaviour seen in the emergency department was used to create a temporal series of data. This was then correlated with a temporal series of lunar luminosity data from the same time period. Crossed correlations in the delay range -7 to +7 days showed coefficient values ranging between -0.102 and +0.034, demonstrating weak correlations without statistical significance. Despite the attractiveness of the popular belief that the moon influences human behaviour, the analysis of our data does not support an association between lunar phases and frequency of violent behaviour. That is, we cannot predict the frequency of cases from a knowledge of lunar luminosity, at least in the period over which our study was performed.