Quantitative techniques employing Markovian processes have been used by a number of investigators to test stratigraphic sections against a null hypothesis of a random succession of sedimentary facies. Rejection of the null hypothesis is used to imply presence of a Markov chain in a succession. However, many of these techniques possess serious statistical flaws arising from the structuring of facies transitions. The result is that presence of preferred facies and regular "cyclic" behavior in sedimentary successions cannot be rigorously supported. Log-linear models, in conjunction with a stepwise procedure for identification of significant transitions, provide a rigorous method for determining the presence and extent of the Markov property in stratigraphic sections. A wide variety of geologic problems that are structured in terms of temporal succession can be examined as a Markov process using log-linear models.