It has been widely perceived that a univariate Gaussian model for evolutionary search can be used to solve separable problems only. This paper explores whether and how the univariate Gaussian model may also be used to solve inseparable problems. The analysis is followed up with experimental tests. The results show that the univariate Gaussian model stipulates no inclination towards separable problems. Further, it is revealed that the model is not only an efficient but also an effective method for solving multimodal inseparable problems. To verify its relative convergence speed, a restart strategy is applied to a univariate Gaussian model (the univariate marginal distribution algorithm) on inseparable problems. The results confirm that the univariate Gaussian model outperforms the five peer algorithms studied in this paper.