A mixed-effects Markov model was developed to describe the probability of ACR20 response (20% improvement in the American College of Rheumatology criteria) as a function of certolizumab pegol exposure, rheumatoid arthritis disease, and placebo response. A model based on the data from placebo-treated subjects was first developed. The structural first-order Markov model described the time course of the transition probabilities between non-responder, responder, and dropout states (i.e., the probabilities for subjects being in each of the three defined states at each study visit conditioned on their state at the previous visit).