Coherency analyzes the trace to trace amplitude similarities recorded by seismic waves. Coherency algorithms have been used to identify the structural or stratigraphic features of an area but the limitations for detecting small-scale features are not known. These limitations become extremely important when interpreting coherency within poorly acquired or processed data sets. In order to obtain a better understanding of the coherency limitations, various synthetic seismic data sets were created. The sensitivity of the coherency algorithms to variations in wave frequency, signal-to-noise ratio and fault throw was investigated. Correlation between the coherency values of a faulted reflector and the known offset shows that coherency has the ability to detect the presence of various scale features that may be previously thought to be below seismic resolution or difficult to discriminate with conventional interpretation methods. Coherency values had a smaller standard deviation and were less sensitive to noise when processed with a temporal window length less than one period. A fault could be detected by coherency when the signal-to-noise ratio was >3. A fault could also be detected as long as the throw-to-wavelength ratio was >5% or two-way traveltime-toperiod >10%. Therefore, this study suggests that coherency has the ability to detect a fault as long as the frequency of the data imaging that fault has a period no greater than one order of magnitude to the traveltime through the fault and that the signal can easily be distinguished from noise. Results from application of the coherency analysis were applied to the characterization of a very deep fault and fracture system imaged by a field seismic data set. A series of reverse and strike-slip faults were detected and mapped. Magnitudes of the throws for these faults were not known, but subtle amplitude anomalies in seismic sections confirmed the coherency analysis. The results of this study suggest that coherency has demonstrated an ability to detect features that would normally beoverlooked using traditional interpretation methods and has many future implications for poorly imaged seismic areas, such as sub-salt.