Models that accurately predict mortality are essential for the managem
ent of forests for sustainable yield. We develop models for mountain a
sh (Eucalyptus regnans) and alpine ash (E. delegatensis) using extensi
ve remeasurement data from Victorian Central Highland forests. The mod
els are based on Cox's proportional hazards regression. The predictive
power of the models is evaluated with data sets withheld from the est
imation procedures. The results suggest that the models will provide a
reliable means of estimating mortality in seeded and planted stands o
f both species on a stand basis.