False Trigger Codes

[liborme]

Apologies if this has been answered elsewhere, i'm quite new to the forum. I've been recording EEG/ERPs with a biosemi 32 channel set up, and all of my files have a series of false triggers at the beginning (at 0 seconds). This is usually quite easy to deal with, but these are sometimes the same trigger codes as the actual triggers and so are less easy to filter out (it is possible, but not ideal!). Can anyone tell me why these occur and how to prevent them?