This paper examines models of the population dynamics of transposable elements when chromosomal sites vary with respect to the effect on fitness of mutations caused by element insertions. Element abundance is assumed to be stabilised solely by the joint results of transposition, excision, and selection against insertional mutations. When there are only two classes of site, selected and neutral, it is hard to find parameter values for which numbers of elements are maintained that match the findings from surveys of Drosophila populations, as elements tend to accumulate at high frequencies at the neutral sites. It is similarly hard to produce realistic equilibria with three classes of site (strongly selected, weakly selected, and neutral), when elements can transpose out of the neutral sites. If transposition from neutral sites is impossible, as might be the case for elements inserted into centric heterochromatin, then realistic equilibria can be generated if there is very weak selection against elements inserted into the majority of non-neutral sites. This model predicts a modest over-representation of elements at the neutral sites. It also predicts that elements should be under-represented on the X chromosome compared with the autosomes, but this is not generally found to be the case. It is concluded that selection against insertional mutations is unlikely to be the major factor involved in the containment of element abundance.