The antiferromagnetic molecular wheel Fe18 of eighteen exchange-coupled Fe III ions has been studied by measurements of the magnetic torque, the magnetization, and the inelastic neutron scattering spectra. The combined data show that the low-temperature magnetism of Fe18 is very accurately described by the Néel-vector tunneling (NVT) scenario, as unfolded by semiclassical theory. In addition, the magnetic torque as a function of applied field exhibits oscillations that reflect the oscillations in the NVT tunnel splitting with field due to quantum phase interference.PACS numbers: 75.50. Xx, 33.15.Kr, 75.10.Jm Although magnetism is a priori quantum mechanical, observation of genuine quantum phenomena such as tunneling and phase interference in magnets is difficult. This is because in ferro-or ferrimagnets, the low-temperature magnetism and dynamics are by and large well described by the magnetization vector M obeying classical equations, although for magnetic molecules (e.g. Mn 12 , Fe 8 ), tunneling of the magnetization and phase interference have been observed [1]. Antiferromagnets, however, exhibit zero magnetization in the ground state, and observation of quantum behavior is even more challenging.Antiferromagnets can be described by the Néel vector n = (M A − M B )/(2M 0 ), with sublattice magnetizations M A , M B of length M 0 (Fig. 1b). In three dimensions, they exhibit long-range Néel order, but domains enable thermally activated or even quantum fluctuations of the Néel vector [2]. Nanosized, single-domain antiferromagnetic (AFM) clusters would provide clean systems to study Néel-vector dynamics, but mono-dispersity is then a key issue. In small particles with weak magnetic anisotropy, the Néel vector can rotate [3,4], while with strong anisotropy it is localized in distinguishable directions, but fluctuates by quantum tunneling, i.e., Néel-vector tunneling (NVT) [5,6,7]. Typically, there are two tunneling paths (Fig. 1c) and interference occurs due to the phase of the wave function. This gives rise to characteristic oscillations in the tunnel splitting as function of applied magnetic field, which would be observable in static measurements such as magnetization [7].Initial attempts to establish NVT with the ferritin protein [8] unfortunately proved controversial owing to polydisperse samples and the presence of uncompensated magnetization [9]. Recent attempts with the AFM molecular wheels CsFe 8 and Fe 10 were encouraging steps forward [10,11], but were criticized with arguments that the tunneling actions S 0 / are too small and the NVT picture only approximately valid [11]. The latter attempts were stimulated by the theoretical prediction of NVT in such wheels [7], and that its observation would be assisted by the monodispersity and crystallinity inherent in molecular compounds, and their resulting well defined structural and magnetic parameters [4,12,13,14].We here present measurements of the magnetic torque (τ = M × B), magnetization, and inelastic neutron scattering (INS) on the AFM molecular wheel [...