We all know that time-ordering depends on the choice of Lorentz frame. So my question is somewhat obvious.... Please give me a hint on where to look up that problem, eg. why the S-matrix theory is covariant. I guess the time-ordering prescription is implicitly defined for each single Lorentz frame (relative times are specific to each frame), and therefore we'll have to adjust the ordering of the operators when applying Lorentz boosts -- thereby rendering the theory covariant in an obvious way. Is that correct?