It is generally accepted that nuclear erythrocytes of some vertebrates are shaped like ellipses. Nevertheless, the formal proof for this has not been presented. Here, the experimental data and theoretical calculations are presented to prove that the shape of nuclear erythrocytes, as a mathematical expectation, is really the ellipse. In addition, allowable limits of deviation of the erythrocytes shape from the ellipse were shown. On the basis of the presented data we have introduced a new parameter (named «ellipsoid coefficient») to evaluate how the real cell shape fits to the correct ellipse. The coefficient is the ratio of the real erythrocyte area (A) to the calculated area (S) obtained by a direct measurement of the erythrocyte major (M) and minor (m) axes. We assume that this parameter can be used in taxonomy of the vertebrate animals having nuclear erythrocytes.