To estimate the time-dependent transmission potential of primary pneumonic plague (PPP), we analyzed historical records from six outbreaks. Based on contact investigation information (source of infection information) of three outbreaks, we generalized the probability density function of the serial interval with a Gamma distribution using maximum likelihood estimations. Furthermore, we used a likelihood-based approach to estimate effective reproduction numbers at time t, Rt, incorporating records from the remaining three outbreaks by assuming independence within unknown contact networks. According to our estimates, the Rt of PPP during the initial phases of each epidemic were roughly in the order of 1.3 (95% confidence interval (CI): 0.0-4.3) in Oakland, 1.4 (0.0-4.6) in Rangoon and 6.5 (0.0-16.0) in Ecuador. The expected values of Rt were shown to slightly exceed unity, even in latter stages. While declining trends in Rt were observed in Oakland and Ecuador, no such trend was observed in Rangoon. The findings suggest that the three outbreaks investigated could have been accompanied by demographic stochasticity. The statistical usefulness of the transformation procedure, even with a small number of recorded cases available, was demonstrated, and the expected responses to bioterrorism using Yersinia pestis were discussed.