Recent reports suggest that autoantibodies directed to aberrantly glycosylated mucins, in particular MUC1 and MUC4, are found in patients with colorectal cancer. There is, however, limited information on the autoantibody levels before clinical diagnosis, and their utility in cancer screening in the general population. In our study, we have generated O-glycosylated synthetic MUC1 and MUC4 peptides in vitro, to mimic cancer-associated glycoforms, and displayed these on microarrays. The assay's performance was tested through an initial screening of serum samples taken from patients at the time of colorectal cancer diagnosis and healthy controls. Subsequently, the selected biomarkers were evaluated in a blinded nested case–control study using stored serum samples from among the 50,640 women randomized to the multimodal arm of the UK Collaborative Trial of Ovarian Cancer Screening (UKCTOCS), where women gave annual blood samples for several years. Cases were 97 postmenopausal women who developed colorectal cancer after recruitment and were age-matched to 97 women without any history of cancer. MUC1-STn and MUC1-Core3 IgG autoantibodies identified cases with 8.2 and 13.4% sensitivity, respectively, at 95% specificity. IgA to MUC4 glycoforms were unable to discriminate between cases and controls in the UKCTOCS sera. Additional analysis was undertaken by combining the data of MUC1-STn and MUC1-Core3 with previously generated data on autoantibodies to p53 peptides, which increased the sensitivity to 32.0% at 95% specificity. These findings suggest that a combination of antibody signatures may have a role as part of a biomarker panel for the early detection of colorectal cancer.