Serum marker evaluation is an easily available prognostic indicator that may help clinicians to discriminate patients with an aggressive disease; there are few and small-sized studies exploring the prognostic role of baseline carcinoembryonic antigen (CEA) values and their variations during first-line therapy, and even fewer data are available for carbohydrate antigen 19–9 (CA 19-9). Our aim was to analyze the role of those prognostic markers to exploit them in daily clinical practice. Data of 892 patients with marker determination before and 3 and/or 6 months during therapy were extracted from two institutional databases. Patients were grouped according to single marker variation as always negative (G0), decreasing (G1), stable (G2), or increasing (G3). We evaluated the progression-free survival (PFS) and the overall survival (OS) of all the patents and correlated them with CEA and CA 19-9 values. A concordance between response to therapy and marker decrease was evident in 50.2 % and in 34.4 % of the patients for CEA and CA 19-9. Patients with low CEA or CA 19-9 baseline values had a longer PFS (15.1 vs. 10.5; 13.6 vs. 10.2 months) and OS (32.0 vs. 22.3; 30.5 vs. 20.1 months). The same results of PFS and OS were obtained by analyzing the data of the four different groups. Multivariate analyses confirmed the independent prognostic role of CEA and CA 19-9. Baseline CEA and CA 19-9 levels and their kinetics demonstrated to be independent prognostic factors. CA 19-9 dosage is not recommended; a possible role of CA 19-9 in patients with negative CEA could be worth further evaluation.