BACKGROUND: Identifying new prognostic factors is important for guiding treatments and preventing metastasis in cancer. Vagal nerve activity may predict prognosis in cancer due to its roles in modulating inflammation, sympathetic activity and oxidative stress. This study tested the relationship between heart rate variability (HRV), a vagal nerve index, and the colon cancer (CC) marker carcinoembryonic antigen (CEA), in an \'historical prospective\' design. METHODS: We examined data of 72 CC patients, without inflammatory or cardiac diseases, of whom 38 had baseline electrocardiograms (ECG) and 12 month CEA levels. We measured HRV (SDNN, RMSSD) from brief archived ECG. Multiple confounders were considered. RESULTS: Controlling for effects of tumor stage and treatment-orientation, baseline HRV predicted CEA levels at 12 months (r=-.43, p=.006). Patients with SDNN<20 ms had significantly higher CEA at 12months than those with SDNN>20 ms. CONCLUSION: These preliminary results showed that higher HRV predicts lower levels of a tumor marker, one year later, independent of confounders. This supports the hypothesized role of vagal activity in tumor modulation. Replication in larger samples is needed.