Abstract: Improved method and circuit are disclosed for processing pulses of a pulse train such that tails of pulses ("A" pulses) which precede a valid true pulse ("T" pulse) to be processed and/or pile-up pulses ("P" pulses) which are piled up on the tails of preceding "T" pulses are eliminated. To eliminate the tails of an "A" pulse a first signal is produced at the occurrence of a "T" pulse. This first signal simulates the tail of the "A" pulse and it is subtracted from the tail piled up "T" pulse. To eliminate a "P" pulse a second signal is produced, when a "P" pulse piled up on the "T" pulse occurs. This second signal simulates the tail of the "T" pulse which tail is cut off at the occurrence of a "P" pulse. This simulated tail is then processed instead of the true tail of the "T" pulse.