This paper has presented a computer model that has been developed to explore the effectiveness of various physical containment strategies. The specific applications that are considered here concern recombinant DNA operations, a technology where technical constraints inhibit the direct monitoring of physical containment effectiveness. Simulations have been used to assess several aspects of recombinant DNA containment, including the effects of using different protocols and host organisms, establishing tradeoffs between physical and biological containment levels, and recognizing the variability introduced by human error and equipment failures. In considering these questions, we have focused on the operator's exposure, although the model that has been developed can also be used to estimate the total release of viable recombinant organisms to the environment. Comparing benchtop protocol simulation results with scale-up protocol results indicates that operator exposure in the former situation may be two or three orders of magnitude lower. A similar difference is apparent between the operator's exposure during a standard scale-up protocol and a protocol that replaces manual sampling with automated sampling. Because the specific activities in a facility influence organism release dramatically, physical containment effectiveness cannot be represented with a single estimate of organisms released per facility per time for each containment level. Results from simulations using E. coli chi 1776 and simulations using hardier organisms indicate that, in a well ventilated facility, the operator's inhalation of viable organisms is essentially unaffected by using the disabled host organism. It should be noted, however, that these simulation results only reflect the differences in the survival of the aerosolized organisms; the relative survival and establishment of organisms after inhalation is not considered.(ABSTRACT TRUNCATED AT 250 WORDS)