Highly significant temporal and spatial seismicity rate changes were found along the Tonga-Kermadec plate boundary. The standard deviate (z) test was used to compare systematically the rates is different periods and different volumes, and to test for uniqueness of anomalies found. The NOAA hypocenter data file showed a pronounced decrease in the number of small-magnitude earthquakes reported after 1969 at the same time as such a reporting decline exists in the worldwide data. Therefore only earthquakes with mb≥4.9 were studied. A search for quiescence before recent mainshocks yielded two uniquely significant precursory anomalies, one missed mainshock and one false alarm. Quiescence started 63 months before the January 1976. M=8.0, Kermadec; and 25 months before the June, 1977, Tonga, M=7.2 earthquakes. The December, 1975, M=7.5, shock was not preceded by a seismicity rate change. The seismic gaps in the Tonga-Kermadec arc with sufficient seismicity for analysis (17¿-22¿ and 31¿-34¿) show constant rates up to the present. Therefore, on the basis of the quiescent hypothesis, earthquakes are not forecast for these segments. The stress drops, Δ&sgr;, of 380 interface events were estimated using the m6/Ms method. The average Δ&sgr; was uniform along the island arc with the exception of two segments with average Δ&sgr; of less than half the normal values. These segments coincided with the two largest recent interface ruptures; howegver, in the northern Tonga the Δ&sgr; was low before the 1975 rupture, whereas in the Kermadec the 1976 aftershocks showed low Δ&sgr;. Asperities could not be identified on the basis of high stress drops. A systematic study of the seismicity rate as a function of space defined large variations along the arc. The areas of highest and lowest seismicity rate correlate with the northern end of the trench and with the Louisville ridge intersection, respectively. However, other large rate contrasts exists between arc segments without obvious tectonic differences. This implies that seismicity doughnut patterns do not reliably define precursors. One segment of outstandingly high seismicity rate (30¿-31¿S) is identified as a major asperity because it contains a large number of m>6 events and a large number of recent mainshock-aftershock sequences and because it stopped the great 1976 rupture. We propose that the only other segment of high seismicity rate (near 20.5 ¿S) also represents a major asperity.