A systematic test on seismic quiescence occurring before largeearthquakes is conducted. For a fixed geographical location, the degree ofclustering in space and time is analysed and the results are testedagainst randomized earthquake catalogs.A gridding technique allows to investigate the entirespatial volume covered by a certain earthquake catalog. The result is a significance K(x,t) for seismic quiescence as a function ofspace and time. A point (x,t) is considered as quiet, if K(x,t) exceeds a threshold value K⁽⁹⁹⁾ such that the nullhypothesis is rejected with a probability of p 99%. Becauseearthquake clusters, like aftershocks and swarm events, generateerroneous quiescence, declustered catalogs are also investigated andthe influence of the clusters is discussed.Applying this method to an earthquake catalog from Armenia,several cases of seismic quiescence before mainshocks are obtained.These quiescence periods occur in the originaldata as well as in the declustered data. Using alarm conditions, itis found that quiescence periods and mainshocks are correlated`better-than-chance'. Thus, the results support the claim thatseismic quiescence makes a contribution to the improvement of seismichazard assessment.