The trigger of the experiment is subdivided into different stages of increasing complexity and has been designed to perform at beam intensities 10 times larger than that of the previous generation experiments.
The first level, called "Level 0", makes use of the trigger hodoscopes, the muon veto and the ring anti-counters to deliver within 500ns a fast trigger signal to the following stage. Its rate is below 1MHz at nominal beam intensity.
At the "Level 1" stage, the Neutral Trigger
decision derived from the information of the photon detector
is available. In a fully pipeline mode, the Neutral Trigger
computes
the energy in the LKr calorimeter, counts the number
of detected peaks in both the x and y projections, computes
the centre-of-gravity of the energy deposition as well as the
position along the beam axis of the 
candidates. It combines also the energy of the hadron calorimeter to
provide the total energy of the event. The expected output rate of the
Neutral Trigger
is less than 2kHz.
The total energy information
is used to reduce the rate at the input of the "Level 2" Charged
Trigger to about 100kHz. At this stage, the Charged Trigger
processes the drift chamber information to compute the vertex
and the invariant mass of 
candidates. The maximum processing time allowed is
100 
s. The output rate of the "Level 2" Charged Trigger is about 1kHz.
Random triggers, calibration triggers as well as downscaled triggers for monitoring purposes can also be provided in parallel by the trigger logic of the experiment.
A "Level 3" stage, not available in 1996, is foreseen to run a complete event reconstruction program on on-line processors so as to keep the total trigger rate below 6kHz.
The event data size is dominated by the LKr calorimeter and depends critically on the suppression of empty channels. The data acquisition system is designed to process 80Mbytes/s at an average rate of 6kHz during 2.4s SPS spills. The total amount of raw data to be put on tape is expected to be 40Tbytes/year.