This article proposes nonnested tests of linear and logarithmic transformations of integrated processes against each other, where the innovations of both series follow autoregressive processes. It is shown that the null distributions of the test statistics for both the linear and logarithmic transformations are nonstandard when the processes have no drift, whereas they are asymptotically normal when the processes have positive drift. Monte Carlo experiments and illustrative empirical examples are provided to show the practical usefulness of the tests in differentiating between linear and logarithmic transformations of a process in finite samples.