Since the collapse of the Dolomieu crater floor at Piton de la Fournaise Volcano (la Réunion) in 2007, hundreds of seismic signals generated by rockfalls have been recorded daily at the Observatoire Volcanologique du Piton de la Fournaise (OVPF). To study rockfall activity over a long period of time, automated methods are required to process the available continuous seismic records. We present a set of automated methods designed to identify, locate, and estimate the volume of rockfalls from their seismic signals. The method used to automatically discriminate seismic signals generated by rockfalls from other common events recorded at OVPF is based on fuzzy sets and has a success rate of 92%. A kurtosis-based automated picking method makes it possible to precisely pick the onset time and the final time of the rockfall-generated seismic signals. We present methods to determine rockfall locations based on these accurate pickings and a surface-wave propagation model computed for each station using a Fast Marching Method. These methods have successfully located directly observed rockfalls with an accuracy of about 100 m. They also make it possible to compute the seismic energy generated by rockfalls, which is then used to retrieve their volume. The methods developed were applied to a data set of 12,422 rockfalls that occurred over a period extending from the collapse of the Dolomieu crater floor in April 2007 to the end of the UnderVolc project in May 2011 to identify the most hazardous areas of the Piton de la Fournaise volcano summit.