Seismic events reported in the bulletins of the two large arrays, LASA and NORSAR, were merged with those from the NEIS bulletin for the period 1970-1977. Using a lower cutoff of mb = 5.8, 510 'main shocks' within the P range of LASA or NORSAR were selected for this period; and various seismicity trends prior to them were investigated. A search for definite foreshocks, based on a significantly short time delay to the main shock, revealed that the true rate of foreshock occurrence was less than 20%. Foreshocks are almost exclusively associated with shallow (h100 km) was demonstrated. For shallow events (<100 km) this variable reveals numerous instances of clustering and spatial-temporal seismic gaps, with little tendency toward a uniformity of behavior prior to main shocks. A statistical test of the validity of seismic precursors was performed for approximately 90 main shock regions which had sufficient seismicity. Using a five-variable vector, (interevent time, interevent distance, magnitude, epicentral distance to main shock, and depth difference relative to main shock) for each event in a 'precursory' time window of 500 days before the main shock and for each event in a 'normal' time window of 500 days before that, the null hypothesis of equal vector means between the two groups was tested. At 90% confidence levels, less than 30% of the main shock regions were thus found to exhibit precursory seismicity changes.