Although a proper derivation of Ohm's law for a particular material should be based on a quantum mechanical treatment of that material, it is interesting to see how far one can proceed on a purely macroscopic basis. It is shown here that for homogeneous, isotropic materials, one can expect an odd voltage-current relation, and that ordinarily Ohm's law will be obeyed as accurately as one wishes over some range of voltage and current. Only the ideas of causality, symmetry, and analyticity are used. Further, if Ohm's law is followed exactly by a material at constant temperature but the resistance is a function of the temperature, and the departure from Ohm's law is due entirely to Joule heating, then it is shown that the odd voltage-current relation still obtains.