We compute starburst models for M82, making use of recent theoretical
tracks of stellar evolution. Detailed comparisons of our models and th
ose of others demonstrate this technique to be quite reliable, with re
latively little change in output parameters as a function of the selec
tion of theoretical tracks or of estimates of the observational charac
teristics of the stars along these tracks. The models are matched to t
he observational constraints for M82 summarized by McLeod et al. The r
ate of star formation and time of observation were thoroughly optimize
d to produce the most favorable fit to the observations, but we still
found that the recently proposed forms for the solar-neighborhood init
ial mass function (IMF) cannot produce starbursts adequate to fit the
observations of this galaxy. We then explored adjustments to the shape
of the IMF to improve the fit to M82. We find (1) the shape of the IM
F for high-mass stars need not be different from that observed locally
; and (2) the most likely modification to the IMF in M82 is that stars
with masses below a few M . form much less commonly than in the solar
neighborhood.