In [1] it was proposed, that the nearest-neighbor distribution P (s) of the spectrum of the Bohr-Mottelson model is similar to the semi-Poisson distribution. We show however, that P (s) of this model differs considerably in many aspects from semi-Poisson. In addition we give an asymptotic formula for P (s) as s → 0, which gives P ′ (0) = π √ 3/2 for the slope at s = 0. This is different not only from the GOE case but also from the semi-Poisson prediction that leads to P ′ (0) = 4The motivation for this comment stems from the impression, that the article "models of intermediate spectral statistics" can easily be misinterpreted in two ways: One might be led to believe that (i) the semi-Poisson distribution is universal, and (ii) the universality class of "intermediate statistics" is as well defined and established as for example the Poisson ensemble or the GOE. In this comment, we will argue, that both statements are wrong.The purpose of [1] is to present models which could constitute a "third" universality class of systems which show so called "intermediate statistics", previously introduced by Shklovskii in [5]. The Poissonian and the Gaussian ensemble (for definiteness, consider orthogonal ensembles only) are considered as the first two universality classes in this list.As in the Poissonian and in the GOE case, where the respective members have common and unique statistical properties, one would expect the same to hold for the models with intermediate statistics. In [1] the authors concentrate on the distribution of nearest neighbor spacings. In the Poissonian case it is given by P (s) = exp(−s), in the GOE case it is close to the well known Wigner surmise P (s) ≈ (π/2) exp(−π/4s 2 ), whereas in the case of the "intermediate statistics