I investigate the influence of a non-zero cosmological constant on the evolution of anisotropy in overdensities that grow by gravitational collapse. The claim that a positive value of LAMBDA might produce stronger asphericities is considered by following the collapse of homogeneous spheroids imbedded in Friedman-Lemaitre backgrounds. It is shown that, although the calculations indeed show this effect, it is small for the values of LAMBDA allowed by the classical cosmological tests. When we do not limit ourselves by constraints arising from the choice of an initial fluctuation spectrum, structures in an open universe (OMEGA0) = 0.1, lamda0 = 0), including the peculiar velocity structure, can be reproduced in a flat Lemaitre universe (OMEGA0 = 0.1, lambda0 = 0.9) for a large part of their evolution. From initial conditions that are not too extreme, these two world models are both able to produce strong anisotropies that easily persist for a Hubble time. This stability of flattening is the only aspect in which these two models differ significantly from the Einstein-de Sitter model. Taking into account the crudeness of the model, together with the fact that observed large-scale structures are not isolated, we conclude that use of this aspect of the dynamics of structure formation is not a promising method of placing tighter constraints on the value of LAMBDA.