Purpose of review: The past couple of years have seen a flurry of publications detailing potential prognostic and predictive biomarkers for colorectal cancer. However, most studies have been underpowered and demonstrate elements of a statistical 'fishing exercise', that is the carrying out of a huge number of analyses to find a hint of importance for a particular marker.
Recent findings: Recently some well powered analyses using separate development and validation datasets and multivariate statistics have been published and these are starting to change attitudes towards such markers.
Summary: Although there are very limited markers in standard clinical practice, we feel that there are now some that are ready to be translated into routine use or that at least deserve some further attention and 'service assessment'.