This paper is concerned with models for the spread of HIV in prisons.
A simple deterministic model for the spread of HIV in a single prison
is considered, a discrete-time stochastic analogue of it is then discu
ssed, and some simulations of epidemics are provided. The model is gen
eralized to a two-prison system in both the deterministic and stochast
ic cases. Finally the possibilities for intervention in a prison are o
utlined and a simple cost-effectiveness analysis in the context of scr
eening and quarantine is undertaken. The computations for the stochast
ic models make use of a new technique for bounding variability in a Ma
rkov chain with a large state space.