We calculate the effective resistivity of a macroscopically disordered two-dimensional conductor consisting of two components in a perpendicular magnetic field. When the two components have equal area fractions, we use a duality theorem to show that the magnetoresistance is nonsaturating and at high fields varies exactly linearly with the magnetic field. At other compositions, an effective-medium calculation leads to a saturating magnetoresistance. We briefly discuss possible connections between these results and magnetoresistance measurements on heavily disordered chalcogenide semiconductors. DOI: 10.1103/PhysRevB.71.201304 PACS number͑s͒: 75.47.De, 61.43.Hv, 72.15.Gd The resistivity of most homogeneous materials ͑metals or semiconductors͒ increases quadratically with magnetic field H at low fields, and generally saturates at sufficiently large H. 1 Exceptions may occur for materials with Fermi surfaces allowing open orbits, or for compensated homogeneous semiconductors, where the resistivity may increase without saturation, usually proportional to H 2 . 1,2 Under some special conditions, the magnetoresistance can be linear in magnetic field. 3 Recently, a remarkably large transverse magnetoresistance ͑TMR͒ has been observed in the doped silver chalcogenides Ag 2+␦ Se and Ag 2+␦ Te. 4,5 In these materials, over the temperature range from 4 to 300 K, the resistivity increases approximately linearly with H up to fields, applied perpendicular to the direction of current flow, as high as 60 T. Moreover, the TMR is especially large and most clearly linear at pressures where the Hall resistivity changes sign. 6 Because of this linearity, these materials may be useful as magnetic field sensors even at megagauss fields.But beyond the possible applications, the origin of the effect remains mysterious. According to conventional theories, such narrow gap semiconductors should have a saturating TMR. Furthermore, since these materials contain no magnetic moments, a spin-mediated mechanism seems unlikely.There are presently two proposed explanations for this quasilinear TMR. The first is a quantum theory of magnetoresistance ͑MR͒. 7 The second proposed mechanism 8 is that this nonsaturating TMR arises from macroscopic sample inhomogeneities. Such inhomogeneities could produce large spatial fluctuations in the conductivity tensor and hence a large TMR, especially at large H. This explanation seems plausible because the chalcogenides probably have a granular microstructure, 6 and hence a spatially varying conductivity.The effective conductivity of media, with a spatially varying conductivity ͑x͒, has been studied since the time of Maxwell, but a relatively few studies have considered the magnetoresistance. 9-18 For a three-dimensional ͑3D͒ medium, the TMR of an isotropic metal does indeed vary linearly in H, when a small volume fraction p Ӷ 1 of inclusions of a different carrier density is added. 15 But the TMR generally does not remain strictly linear at higher concentrations of p. If the inclusions are strictly insulatin...