Abstract Background Circulating tumor cells (CTCs) represent one of the most interesting target in improving diagnosis, prognosis and treatment. Herein we evaluate the possibility of using an emo-cytometric approach on the evaluation of the heterogeneous population of CTCs to improve personalized metastatic risk assessment. We benchmarked ex vivo behavior of distinct subsets of circulating colon tumor cells with correspondent clinical behavior of patients from which we isolated CTCs. Methods Isolation and CTC expansion were performed by a gradient protocol. In vitro characterization was determined by flow cytometry, immunofluorescence, western blotting and proteomic profiling. Cell sorter was performed with immunomagnetic beads. Confocal microscopy was used to evaluate tissue sections. Kaplan Mayer curves was cared for through Medcalc program. Results We collected heterogeneous CTCs, derived from the whole blood of seven patients affected by colon cancer, expressing CD133posCD45neg (5 ± 1) and (2 ± 1) and CK20posCD45neg of (29 ± 3) (11 ± 1) cells/ml in Dukes D and A stage respectively. Proliferation rate of 57 ± 16 %, expression for CXCR4pos of 18 ± 7 % and detectable levels of IL-6, IL-8 and SDF-1 cytokines in conditioned culture medium characterized short-time expanded–CTCs (eCTCs). ECTCs organized in tumor sphere were CD45negCD133pos while in adhesion were CXCR4posCK20pos. These two subsets were separately injected in mice. The first group of xenografts developed superficial lesions within 2 weeks. In the second group, in absence of growing tumour, the survival of injected eCTCs was monitored through SDF-1 serum levels detection. The detection of human cancer cells expressing CK20, in mice tissues sections, suggested a different biological behaviour of injected eCTC-subsets: tumorigenic for the first and disseminating for the second. The benchmarking of the experimental data with the clinical course highlights that patients with prevalence of circulating cancer stem cells (CD45negCD133pos) have a lower overall survival. Conversely, patients with prevalence of circulating differentiated cells (CXCR4posCK20pos) have a low disease-free survival. Conclusion On the basis of the heterogeneous composition and despite the low number of CTCs, it was possible to distinguish two subgroups of CTCs, suggesting a different clinical outcome. CTC-subsets detailing is useful to better define the metastatic–risk personalized score thus improving disease management and reducing patient care cost.