The retrieval of information con veyed in data recorded by seis mic arrays plays a key role in seismology and geophysical exploration. Accurate localization and reli able detection of seismic events are major tasks in seis mic monitoring systems. The nonstationarity, low signal-to noise ratio (SNR), and weak signal coherence of seismic data remain challenging issues for signal processing algorithms. The maximum likelihood (ML) approach that performs well in such critical conditions is one of the best solutions for simultaneous detec tion and localization of seismic events. This article will discuss the methodology of ML for estimation and detection of seismic data and its extension to geoacoustic model selection.