Controlling an emerging communicable disease requires prompt adoption of measures such as quarantine. Assessment of the efficacy of these measures must be rapid as well. In this paper, the authors present a framework to monitor the efficacy of control measures in real time. Bayesian estimation of the reproduction number R (mean number of cases generated by a single infectious person) during an outbreak allows them to judge rapidly whether the epidemic is under control (R < 1). Only counts and time of onset of symptoms, plus tracing information from a subset of cases, are required. Markov chain Monte Carlo and Monte Carlo sampling are used to infer the temporal pattern of R up to the last observation. The operating characteristics of the method are investigated in a simulation study of severe acute respiratory syndrome-like outbreaks. In this particular setting, control measures lacking efficacy (R > or = 1.1) could be detected after 2 weeks in at least 70% of the epidemics, with less than a 5% probability of a wrong conclusion. When control measures are efficacious (R = 0.5), this situation may be evidenced in 68% of the epidemics after 2 weeks and 92% of the epidemics after 3 weeks, with less than a 5% probability of a wrong conclusion.