Mathematical models are developed for the chemotherapy of AIDS. The mo
dels are systems of differential equations describing the interaction
of the HIV infected immune system with AZT chemotherapy. The models pr
oduce the three types of qualitative clinical behavior: an uninfected
steady state, an infected steady state (latency) and a progression to
RIDS state. The effect of treatment is to perturb the system from prog
ression to AIDS back to latency. Simulation of treatment schedules is
provided for the consideration of treatment regimes. The following iss
ues of chemotherapy are addressed: (i) daily frequency of treatment, (
ii) early versus late initiation of treatment and (iii) intermittent t
reatment with intervals of no treatment. The simulations suggest the f
ollowing properties of AZT chemotherapy: (i) the daily period of treat
ment does not affect the outcome of the treatment, (ii) treatment shou
ld not begin until after the final decline of T cells begins (not unti
l the T cell population falls below approximately 300 mm(-3)) and then
, it should be administered immediately and (iii) a possible strategy
for treatment which may cope with side effects and/or resistance, is t
o treat intermittently with chemotherapy followed by interruptions in
the treatment during which either a different drug or no treatment is
administered. These properties are revealed in the simulations, as the
model equations incorporate AZT chemotherapy as a weakly effective tr
eatment process; We incorporate into the model the fact that AZT treat
ment does not eliminate HIV, but only restrains its progress. The math
ematical model, although greatly simplified as a description of an ext
remely complex process, offers a means to pose hypotheses concerning t
reatment protocols, simulate alternative strategies and guide the qual
itative understanding of AIDS chemotherapy.