ABSTRACT Kinematic studies have produced accurate measurements of the total dark matter mass and mean dark matter density within the optical extent of galaxies for large samples of objects. Here we consider theoretical predictions for the latter quantity, $\bar{\rho }_{\rm dm}$, measured within the isophotal radius R23.5, for isolated haloes with universal density profiles. Through a combination of empirical scaling relations, we show that $\bar{\rho }_{\rm dm}$ is expected to depend weakly on halo mass and strongly on redshift. When galaxy haloes fall into larger groups or clusters, they become tidally stripped, reducing their total dark matter mass, but this process is expected to preserve central density until an object is close to disruption. We confirm this with collisonless simulations of cluster formation, finding that subhaloes have values of $\bar{\rho }_{\rm dm}$ close to the analytic predictions. This suggests that $\bar{\rho }_{\rm dm}$ may be a useful indicator of infall redshift on to the cluster. We test this hypothesis with data from the SHIVir survey, which covers a reasonable fraction of the Virgo cluster. We find that galaxies with high $\bar{\rho }_{\rm dm}$ do indeed trace the densest regions of the cluster, with a few notable exceptions. Samples selected by environment have higher densities at a significance of 3.5–4σ, while samples selected by density are more clustered at 3–3.5σ significance. We conclude that halo density can be a powerful tracer of the assembly history of clusters and their member galaxies.