Iron(II) spin crossover (SCO) materials have been widely studied as molecular switches with a wide variety of potential applications, including as displays, sensors, actuators or memory components. Most SCO materials have been either monometallic or polymeric, and it is only relatively recently that chemists have really started to focus on linking multiple metal centres together within the one, discrete, molecule in an effort to enhance the SCO properties, such as abrupt, hysteretic, and multistep switching, as well as the potential for quantum cellular automata, whilst still being readily amenable to characterisation. Here we present a review of the ligand designs of the last two decades that have led to self assembly of discrete di- to poly-nuclear iron(II) complexes of helicate, cage, cube, and other supramolecular architectures with rich SCO activity, and to an increased focus on host–guest interactions. Analysis of selected octahedral distortion parameters (Σ, CShM) reveals interesting differences between these structural types, for example that the iron(II) centres in grids are generally significantly more distorted than those in squares or cages, yet are still SCO-active. Of the 127 complexes reviewed (79 published 2012–Feb. 2018), 54% are dinuclear, 10% trinuclear, 31% tetranuclear, and the remaining 5% are penta, hexa and octanuclear. Of the 93 designer ligands utilised in these polynuclear architectures: 60 feature azoles; 55 provide all donors to the Fe(II) centres (no co-ligands coordinated) and form exclusively 5-membered chelate rings via either bidentate azole-imine/pyridine or tridentate heterocycle-imine/amine/thioether/pyridine-heterocycle binding pockets.