The occurrence of triggered and nontriggered substorm are examined in light of current interest in such issues as substorm identification, IMF By variations, and potentially undetected small-scale solar wind perturbation. Global substorms are identified using a sudden, persistent decrease in the AL index. The onset of this global expansion is taken to be the time of the Pi 2 burst nearest in time to the beginning of the AL decrease. IMF triggers were identified both subjectively through visual scanning of the data and automatically with a computer algorithm. Both northward turnings of the IMF Bz and decreases in the amplitude of the By component were considered as possible triggers. Two different solar wind monitors were used in the investigation: IMP-8 in a circular orbit with a distance 12~35 Re to the Earth-Sun line and ISEE-2 in an elliptical orbit with a distance only 5~10 Re to the Earth-Sun line. The IMP-8 results show that the triggering probability does not depend on the distance of the monitor from the Earth-Sun line in the range 12--35 Re. The ISEE dataset shows that closer than 12 Re the triggering probability is the same as it is in the IMP-8 data set. Thus there appears to be no dependence of triggering on the location of the monitor provided it is within 35 Re of the Earth. We also demonstrate that including the By component does not significantly increase the probability of substorm triggering. Approximately 60% of all substorms appear to be triggered. Of the 40% for which we could not identify a trigger, 10% occurred while the IMF was northward. The data suggest that substorm onset is a consequence of an internal magnetospheric instability that is highly sensitive to changes in magnetospheric convection induced by a sudden change in the IMF, but that these changes are not always necessary.