Quantifying transitions in seismic activity related to wastewater injection is an important step for accurate seismic‐hazard assessment. It is also a challenging task due to the uncertainty in the relationship between injections and transitions; however, a consistent statistical analysis of instrumental seismic records allows for detecting and quantifying induced seismicity. In this study, we develop a statistical method for modeling a seismic sequence involving non‐stationary‐induced seismicity. It is composed of two steps: first, we select a model for the integrated seismicity (i.e., natural and induced) within the framework of the epidemic‐type aftershock sequence via the Bayesian Model Comparison section. Second, we perform Bayesian inference within that model to assess the seismic activity and associated parameters. The method is applied to the analysis of the events from Oklahoma, demonstrating that it is able to provide a consistent representation of the occurrence of the dataset. Results show that the overall seismic rate (including mainshock and aftershock events) for events with local magnitudes (ML) above 2.5 has been escalated by a factor of more than 100, from 0.05 to more than five events per day, between January 1975 and August 2014. For this overall increase, the contribution of the mainshock events is estimated to be ∼56%. Assuming the b‐value of the Gutenberg–Richter law is 1.0, the probability of exceeding ML 5.0 in a 2‐month period is predicted to have increased from about 0.05 to more than 0.5 during the study period. A sensitivity analysis is presented to show how the probabilistic inference is affected by the assumed b‐value and the assumed maximum event magnitude. The proposed method can provide a statistical basis for quantitatively assessing the process of induced seismicity. In addition, it can be employed as a decision‐support tool to identify areas with increasing levels of hazard and to guide strategies for risk mitigation.