An outstanding question about the underdoped cuprates concerns the true nature of their Fermi surface which appears as a set of disconnected arcs. Theoretical models have proposed two distinct possibilities: ͑1͒ each arc is the observable part of a partially hidden closed pocket and ͑2͒ each arc is open, truncated at its apparent ends. We show that measurements of the variation in the interlayer resistance with the direction of a magnetic field parallel to the layers can qualitatively distinguish closed pockets from open arcs. This is possible because the field can be oriented such that all electrons on arcs encounter a large Lorentz force and resulting magnetoresistance whereas some electrons on pockets escape the effect by moving parallel to the field. DOI: 10.1103/PhysRevB.82.172510 PACS number͑s͒: 74.25.fc, 74.25.Jb, 74.72.Kf, 75.47.Np The Fermi surface ͑FS͒ of underdoped cuprates in the pseudogap state appears, in electronic spectrum measurements, as four short arcs near diagonals of the Brillouin zone.1-8 These arcs neither close back on themselves nor terminate at zone boundaries, which are the only possibilities for a conventional FS, but rather end abruptly within the zone interior. According to some theoretical pictures, [9][10][11][12][13] each apparently open spectral arc is just the observable segment of a closed Fermi-surface pocket ͑the missing side of the pocket is claimed to be present but undetected because of its lower spectral weight͒. In contrast, others propose that truly open arcs, without any closed pockets, comprise the FS. [14][15][16] In this Brief Report, we show that the interlayer magnetoresistance ͑IMR͒ is qualitatively different for closed pockets and open arcs. Hence, the IMR measurements we propose should be able to rule out a whole class of theoretical models for the pseudogap state.Though quasiparticle peaks on the arcs are broad in zero magnetic field, the observation of quantum oscillations ͑QOs͒ in underdoped cuprates [17][18][19][20][21] indicates that sharp quasiparticles are present in high fields. Based on their frequency, the oscillations may be plausibly attributed to quasiparticles on the spectral arcs 22 but either closed pockets or open arcs 23 can accommodate QOs. To elucidate the connection between QOs and the nature of the spectral arcs we need a complementary probe, one that accesses the high-field phase where QOs are seen and determines whether the quasiparticles more likely live on a closed or open FS.The dependence of the IMR on the direction of the magnetic field has proven to be a powerful probe of Fermisurface properties in overdoped cuprates. [24][25][26] We have previously proposed that it can be used to map the anisotropy of a weak pseudogap. 27 Significant IMR effects require a magnetic field strong enough that the cyclotron frequency C is of order the scattering rate −1 , the same condition needed for QOs. 28 When the field B is in the conducting layers, only quasiparticles moving parallel to B, which feel no Lorentz force, avoid a large classica...