Abstract To investigate possible indicators of critical point behavior prior to rock failure, the statistical properties of pre-failure damage were analyzed based on acoustic emission events (AE) recorded during the catastrophic fracture of typical rock samples under differential compression. AEs were monitored using a high-speed 32-channel waveform recording system. Time-dependent statistics, including the energy release rate, b -value of the magnitude–frequency distribution, fractal dimension and spatial correlation length (SCL) of the AE hypocenters were calculated for each data set. Each parameter is a function of the time-to-failure and thus can be treated as an indicator of the critical point. It is clear that the pre-failure damage evolution prior to catastrophic failures in several common rock-types is generally characterized by: 1) accelerated energy release, 2) a decrease in fractal dimension and SCL with a subsequent precursory increase, and 3) a decrease in b -value from ∼ 1.5 to ∼ 0.5 for hard rocks, and from ∼ 1.1 to ∼ 0.8 for soft rocks such S–C cataclasite. However, each parameter also reveals more complicated temporal evolution due to either the heterogeneity of the rock mass or the micro-mechanics of shear fracturing. This confirms the potential importance of integrated analysis of two or more parameters for successfully predicting the critical point. The decreasing b -value and increasing energy release may prove meaningful for intermediate-term prediction, while the precursory increase in fractal dimension and SCL may facilitate short-term prediction.