Accurately locating microseismic events can be used to monitor fluid migration during reservoir production and hydrofracturing operations. This is usually done with sparse networks of seismic sensors located in boreholes or with denser surface arrays. The data used for microearthquake monitoring are corrupted by noise, which reduces the signal-to-noise ratio to values as low as 0.1. Monitoring methods based on traveltime picking of various wave modes (P or S) cannot deal with this level of noise and require extensive user interaction. An alternative class of methods uses time reversal to focus microearthquake information at the source position. These methods can handle noisier signals because they do not rely on event picking or recorded seismograms, but the methods are also costlier to run and still require picking locations where wavefields focus. We advocate the time-reversal technique within the general framework of Bayesian inversion. Given an assumption about the possible locations of microearthquakes, we use recorded data to evaluate the feasibilityof microearthquakes occurring at various locations in the earth. The method takes into account imaging imperfections caused by unknown components of the model or acquisition array aperture. We simulate wavefields corresponding to possible sources distributed in the model and evaluate their match with the wavefield reconstructed from the synthetic data recorded in the model. In this regard, the method operates like a pattern-recognition procedure and can exploit a wide variety of techniques designed for this purpose. We use simple crosscorrelation between simulated and reconstructed wavefields to take advantage of the speed and robustness of this technique. The wavefields reconstructed at various locations are used over time to scan the wavefield constructed from field data. Thus, our method can identify not only the position of the microearthquakes but also their origin times as well as the orientations of the fault planes characterizing the seismic sources. Maps of probability are the final outcome of this automated process, indicating the confidence of microearthquake occurrence at various positions and times.