High-throughput single-cell analysis is a challenging target for implementing advanced biomedical applications. An excellent candidate for this aim is label-free tomographic phase microscopy. However, in-line tomography is very difficult to be implemented in practice, as it requires complex setup for rotating the sample and/or illuminate the cell along numerous directions . We exploit random rolling of cells while they are flowing along a microfluidic channel demonstrating that it is possible to obtain in-line phase-contrast tomography by adopting strategies for intelligent wavefronts analysis thus obtaining complete retrieval of both 3D-position and orientation of rotating cells . Thus, by numerical wavefront analysis a-priori knowledge of such information is no longer needed. This approach makes continuos-flow cyto-tomography suitable for practical operation in real-world, single-cell analysis and with substantial simplification of the optical system avoiding any mechanical/optical scanning of light source. Demonstration is given for different classes of biosamples, red-blood-cells (RBCs), diatom algae and fibroblast cells . Accurate characterization of each type of cells is reported despite their very different nature and materials content, thus showing the proposed method can be extended, by adopting two alternate strategies of wavefront-analysis, to many classes of cells. In particular, for RBCs we furnish important parameters as 3D morphology, Corpuscular Hemoglobin (CH), Volume (V), and refractive index (RI) for each single cell in the total population . This could open a new route in blood disease diagnosis, for example for the isolation and characterization of "foreign" cells in the blood stream, the so called Circulating Tumor Cells (CTC), early manifestation of metastasis.