Current terrestrial gravitational-wave detectors operate at frequencies above 10Hz. There is strong astrophysical motivation to construct low-frequency gravitational-wave detectors capable of observing 10mHz-10Hz signals. While space-based detectors provide one means of achieving this end, one may also consider terretrial detectors. However, there are numerous technological challenges. In particular, it is difficult to isolate test masses so that they are both seismically isolated and freely falling under the influence of gravity at mHz frequencies. We investigate the challenges of low-frequency suspension in a hypothetical terrestrial detector. As a case study, we consider a Magnetically Assisted Gravitational-wave Pendulum Intorsion (MAGPI) suspension design. We construct a noise budget to estimate some of the required specifications. In doing so, we identify what are likely to be a number of generic limiting noise sources for terrestrial mHz gravitational- wave suspension systems (as well as some peculiar to the MAGPI design). We highlight significant though not necessarily insurmountable experimental challenges.