Studies on linkage disequilibrium (LD) are important in mapping disease genes. A novel statistical method, the multi-order Markov chain model has been recently developed to quantify the complexity level of multilocus LD patterns among single nucleotide polymorphism markers (Kim et al., 2008). In this study, mathematical relationships between two types of LD measures are derived to understand the Markov chain model parameters in terms of conventional LD measures. Statistical sample properties of the Markov chain order estimates are investigated by simulations. Two published data sets are reanalyzed to illustrate the proposed approach.