Testing the zoo examine out on MC files requires running on L3 trigger simulator banks. These aren't always present. This web page describes how to run the zoo module on MC files that don't have the trigger banks present.
There is also a page of release info which contains caveats for running the trigger simulator with an online trigger list on some releases.
The Tevatron collides protons and antiprotons at about 7 MHz. The triggers' job is to reduce that rate to about 50 Hz, and do it intelligently. This means looking at every single event and deciding, at a rate of 7 MHz, to keep or not to keep the data. The DZERO trigger does this in three stages (as do most modern triggers): L1, L2, and L3. L1 runs in real-time. It does the most basic cuts (does this event contain energy in the calorimeter? How about tracks? Are they even close to each other?). L2 doesn't run in real-time -- it just keeps up on average (as does L3). Its input rate is 10 kHz, and it trys to find jets in the calorimeter, looks for EM objects that are matched to tracks, etc. In L3, which is a farm of PCs running Linux (and which UW works on) a fairly sophisticated decision can be made. This is also the first time that complete data readout is present (some detectors have fast-outs and give low-resolution data to earlier stages of the trigger). Its input rate is 1 kHz, and its output, about 50 Hz, is written to tape for later, offline, analysis.
Note -- there is something called "giving up" in a trigger. That is, you can't make up your mind, but the data is coming too fast, so you just cut out some random number of events. This is called prescaling a trigger.If a trigger has a prescale of 50, and its raw rate is 50 Hz, then after the prescale its rate will be 1 Hz. Under normal circumstances, you want to avoid this situation as you've not made an intelligent decision about the event!
A trigger list is just a series of requirements (1 jet > 20 GeV, 1 isolated electron with 10 GeV and a matching track). In the old days an experiment would have, say, 32 separate triggers. Collectively, they were known as a trigger word, and each trigger was associated with a trigger bit. Even though we don't hardware anything any more, the terminology is still with us. Each trigger list is composed of a collection of trigger bits. L1 and L2 can have up to 128 trigger bits. This limit is defined by bus widths in hardware, and is difficult to get around. In L3 everything is done in software. Currently 256 is the software limit.
There is actually quite a bit of information out there on trigger lists (considering it is so important to the experiment...)