Background Cell-cell communication and adhesion are essential for the compaction process of early stage embryos. The aim of this study was to develop a non-invasive objective calculation system of embryo compaction in order to test the hypothesis that embryos with a larger mean contact surface result in a higher live birth rate compared to embryos with a lower mean contact surface. Methods Multilevel images of 474 embryos transferred on day 3 were evaluated by the Cellify software. This software calculates the contact surfaces between the blastomeres. The primary outcome of this study was live birth. An ideal range of contact surface was determined and the positive and negative predictive value, the sensitivity, the specificity and the area under the curve for this new characteristic were calculated. Results In total, 115 (24%) transferred embryos resulted in a live birth. Selection of an embryo for transfer on its mean contact surface could predict live birth with a high sensitivity (80%) and high negative predicting value (83%) but with a low positive predictive value (27%), a low specificity (31%) and low area under the ROC curve (0.56). The mean contact surface of embryos cultured in a single medium was significantly higher compared to the mean contact surface of embryos cultured in a sequential medium (p = 0.0003). Conclusions Neither the mean contact surface nor the number of contact surfaces of a day 3 embryo had an additional value in the prediction of live birth. The type of culture medium, however, had an impact on the contact surface of an embryo. Embryos cultured in a single medium had a significant larger contact surface compared to embryos cultured in the sequential medium.