The density of dark matter near the Sun, ρ_{DM, ⊙}, is important for experiments hunting for dark matter particles in the laboratory, and for constraining the local shape of the Milky Way’s dark matter halo. Estimates to date have typically assumed that the Milky Way’s stellar disc is axisymmetric and in a steady-state. Yet the Milky Way disc is neither, exhibiting prominent spiral arms and a bar, and vertical and radial oscillations. We assess the impact of these assumptions on determinations of ρ_{DM, ⊙} by applying a free-form, steady-state, Jeans method to two different N-body simulations of Milky Way-like galaxies. In one, the galaxy has experienced an ancient major merger, similar to the hypothesized Gaia–Sausage–Enceladus; in the other, the galaxy is perturbed more recently by the repeated passage and slow merger of a Sagittarius-like dwarf galaxy. We assess the impact of each of the terms in the Jeans–Poisson equations on our ability to correctly extract ρ_{DM, ⊙} from the simulated data. We find that common approximations employed in the literature – axisymmetry and a locally flat rotation curve – can lead to significant systematic errors of up to a factor ∼1.5 in the recovered surface mass density ∼2 kpc above the disc plane, implying a fractional error on ρ_{DM, ⊙} of the order of unity. However, once we add in the tilt term and the rotation curve term in our models, we obtain an unbiased estimate of ρ_{DM, ⊙}, consistent with the true value within our 95 per cent confidence intervals for realistic 20 per cent uncertainties on the baryonic surface density of the disc. Other terms – the axial tilt, 2nd Poisson and time-dependent terms – contribute less than 10 per cent to ρ_{DM, ⊙} (given current data) and can be safely neglected for now. In the future, as more data become available, these terms will need to be included in the analysis.