Microseismic multiplets, which are groups of events showing highly similar waveform despite different origin time and magnitude, can be effectively used to determine relative location of the hypocenters with high accuracy. The authors clustered multiplets from microseismic events collected during a hydraulic stimulation in the Deep Heat Mining Project at Basel, Switzerland in 2006. The similarity of waveform was quantitatively evaluated in the frequency domain using the magnitude square coherence function, and this made it possible to cluster the events with different criteria which correlate to different physical phenomena associated with shear slip on fractures. We then interpreted the behavior of each multiplet cluster from the orientation and size of the multiplet seismic structure, fault plane solution, source radius, and hydraulic record. Multiplets which correlate to (a) identical shear slip on a macroscopic single fracture or thin fracture network, (b) repeating slip within a microscopic fracture, and (c) extension of rupture zone on a microscopic fracture, are identified and their response to the stimulation was also interpreted.