Many electron and X-ray microscopy applications place conflicting demands on detector subsystems. These applications seek to maximise detector throughput at high count-rates at the same time as minimising pulse pile-up and maintaining accurate energy resolution. Traditional approaches to digital pulse processing typically require the user to trade-off degraded energy resolution against detector dead time.This paper presents a digital pulse processing technology for processing the digitised output of Xray detectors. Using a multi-stage, nonlinear digital signal processing algorithm this method is able to decode pile-up events in real-time, accurately characterizing the number, time-of-arrival and energy of all X-ray pulses in the digitised detector output. This digital pulse processing technology enables high count-rate operation with little dead-time, improved energy resolution and very high detector throughput, dramatically improving the performance of many electron and X-ray microscopy applications.The performance of the nonlinear digital pulse processing algorithm has been evaluated over a wide range of input count-rates using both a digital nuclear pulse generator and a range of X-ray calibration sources. Implemented on a PCI based computer card, in a Virtex-IV FPGA, the digital pulse processing algorithm is able to decode pulse pile-up in real-time and is compatible with a wide range of X-ray detection systems.Pulse pile-up is evident in the electrical output signal of an X-ray detector when events arrive within such a small time difference that the detector cannot recover sufficiently from the first radiation event to accurately detect subsequent events. Consequently the output signal, which ideally would consist of discrete pulses each corresponding to the incidence of a single radiation event, exhibits a waveform in which individual pulses overlap or "pile-up". In X-ray microscopy systems the effects of pulse pile-up include reduced throughput, system dead-time, and degradation of energy resolution, which can significantly inhibit overall performance [1], [2].In order to fully characterise the output signal from the X-ray measurement process it is necessary to estimate the number, amplitude and time-of-arrival of incoming radiation events along with the impulse response of the detection process. The real-time pulse pile-up recovery algorithm requires a number of steps to accurately estimate the number, time of arrival, and energy of all radiation events in the digitised detector data. These steps are illustrated in Fig. 1 and explained below.System Characterisation determines the system attributes unique to the particular X-ray detector and is required for accurate pulse localisation and identification.Pulse localisation is performed on the input data stream in order to accurately determine the timeof-arrival of each radiation event.Pulse identification uses the digitised detector output, the system characterisation data and the pulse localisation data to determine the energy of each pulse in the digitis...