This study presents an agglomerative hierarchical clustering-based strategy using Shared Nearest Neighbours and multiple dissimilarity measures to identify typical daily electricity usage profiles of university library buildings. The proposed strategy takes the advantages of three dissimilarity measures (i.e. Euclidean distance, Pearson distance and Chebyshev distance) to calculate the difference between daily electricity usage profiles. Two-year hourly electricity usage data collected from two different university library buildings were employed to evaluate the performance of this strategy. It was shown that this strategy, which considered both magnitude dissimilarity and variation dissimilarity simultaneously, can identify more informative typical daily electricity usage profiles, in comparison with other twelve clustering-based strategies which used a single dissimilarity measure. Some interesting information related to building energy usage behaviours was also discovered with the help of visualisation techniques. Additional or hidden information discovered using this strategy can potentially be useful for fault detection and diagnosis and performance enhancement of library buildings.