The basis for prognostic prediction after surgery for colorectal cancer remains the various pathological staging systems based on that of Dukes. Serum prognostic markers have not shown significant independent prognostic power compared with these predictive tools. Much energy has been expended in examining the ability of serum markers to predict recurrent tumour prior to the onset of symptoms. Carcinoembryonic antigen (CEA) has been a particular subject of attention, and has been widely, though variably, advocated as a useful predictor in these circumstances. It has been estimated that around half a million Americans are presently undergoing regular postoperative CEA monitoring to this end. Controversy continues regarding the therapeutic utility of such monitoring. This may be resolved when the results of the only randomised trial in the field are published in the near future. No other serum marker, nor any combination of markers, has been shown clearly to be superior to CEA as a predictor of recurrent tumour.