A novel synthesis of the title compound, C₂S₃N₃ (1) is reported. X- and K-band EPR spectra on dilute solutions of 1 indicate delocalisation of the unpaired spin density over both heterocyclic rings in agreement with DFT calculations. An XRPD study indicates that it crystallises in two morphologies with both phases formed during vacuum sublimation. The XRPD studies indicate that on cooling below 230 K, only the triclinic phase (P) becomes detectable, whereas on warming above 320 K, just the monoclinic phase (P2₁/c) becomes observed. The crystal structure of the monoclinic phase has been examined by variable temperature single crystal X-ray diffraction in the region 300–225 K and reveals a regular π-stacked structure. A crystal structure of the triclinic phase is reported at 150 K and exhibits a dimeric π-stacked motif. Susceptibility measurements show that the monoclinic phase is paramagnetic whereas the triclinic phase is diamagnetic. This radical exhibits thermal hysteresis with a wide range of bistability; EPR and magnetic susceptibility measurements indicate Tc_↓ = 234 K, and Tc_↑ = 317 K. The magnetic behaviour of the monoclinic phase is consistent with strong antiferromagnetic exchange interactions between open shell doublet states (J = −320 K) along the π-stacking direction, although significant inter-stack interactions are required to model the data adequately. In contrast the dimeric phase is essentially diamagnetic, with the residual paramagnetism indicating a very large singlet–triplet separation (|2J| > 2000 K). The magnetic exchange interactions in both phases are probed through a series of DFT calculations using the broken-symmetry approach. These confirm the presence of strong magnetic exchange interactions along the π-stacking direction in the high temperature phase (2J = −182 K), but with additional interstack interactions which are an order of magnitude smaller. Calculations on the triclinic phase indicate that it is best considered as a dimer with an open-shell singlet state with a very large singlet–triplet separation (2J = −2657 K). The magnitude of J for both phases from theory and experiment are in good agreement. The origin of the thermal hysteresis is attributed to the presence of two energetically similar structures which have a low energy barrier to interconversion. The thermodynamic parameters associated with the interconversion process have been probed by DSC studies. It confirms the first order nature of the transition with Tc_↓ = 232.3 K (ΔH_↓ = 1.41 kJ mol⁻¹, ΔS_↓ = 6.0 J mol⁻¹ K⁻¹) and Tc_↑ = 320.5 K (ΔH_↑ = 1.86 kJ mol⁻¹, ΔS_↑ = 5.8 J mol⁻¹ K⁻¹).