More than 330 earthquakes, with maximum intensity (Im) greater than II and for which both location and size can be determined, are known to have occurred within New York state between 1720 and 1980. The instrumental data on seismicity (1970 to present) indicate that New York State can be subdivided into three regions of relatively high seismic activity separated by a large central region of little or no activity (i.e., more than 90% of activity occurs within 60% of the state’s area). The historical data suggest that, in general, the same areas have been active to varying degrees since 1720. There has never been a case of clear surface faulting detected with any of the earthquakes, and in only a few cases has it been possible to correlate earthquakes with known faults, even though all events are suspected to be shallow (h < 15 km). The slope (b-value) in the cumulative frequency-intensity plot, based on data for the whole state since 1900, for events with Im from IV through VII, is 0.51. The b-values for the four subregions of the state are less well determined but are in the range 0.5 to 0.55. The frequency-intensity data imply that the record for events greater than III is complete since 1900 throughout the state. Therefore, a dramatic increase noted in the number of IV and larger events since 1900 appears to be real and not merely the result of improvement in detection capability. In terms of total seismic energy release, which is determined exclusively by the largest events, the occurrence of which are known back to 1730, the seismic activity shows large systematic secular variations in the form of several peaks and troughs. For the period 1900 through 1980, the average calculated recurrence time for an earthquake of Im ≥ VIII is about 100 years for the total state. Similar estimates for Im ≥ IX and X are 376 and 1360 years, respectively. Earthquakes greater than VIII have not been observed. However, from the statistical evidence alone, there is no reason to believe that such events cannot occur in New York state, because their average repeat times are considerably longer than the available historical record. Also, because of the large secular variation in seismic activity, the above estimate for the repeat times, based on data from 1900 through 1980, may not adequately represent average values for a longer period or predict activity into the immediate future. The present understanding of tectonics is also insufficient to preclude the possible occurrence of events greater than VIII.