Recent treatment advances suggest that AIDS may be changing from a rap
idly lethal, acute disease into a chronic disease process with lengthy
periods of remission. Stochastic compartment models can describe AIDS
progression, remission, and overall survival. Such models can combine
multiple data sources to estimate transitions through intermediate, u
nobserved disease states. Age/time AIDS mortality may be combined with
age/time AIDS incidence to generate an ''observed'' health event dist
ribution, using maximum likelihood estimation. The models can project
the future behavior of AIDS in an aging population. Transitions among
intermediate disease states (i.e., components of the disease's ''natur
al'' and ''treatment-altered'' history) can be evaluated as time-varyi
ng functions that reflect treatment efficacy For example, the introduc
tion of protease inhibitors to therapy with two nucleoside analogues s
ignificantly slows transitions between disease states. Persons infecte
d, but successfully treated, may develop organ damage, neurological di
sorders, or cancer.