A signal molecule, such as a hormone, interacts with receptor molecules triggering a cascade of downstream events which we call its pathway. Xenobiotic chemicals can alter normal biological functions by interacting with receptors in ways that mimic natural signaling molecules. A new way to test xenobiotic chemicals for interference along these pathways is to use simple cell-based assays. These tests are considerably cheaper and far less time-consuming than traditional animal-based tests (a very important consideration when we think of the tens of thousands of chemicals to which we are exposed). However, the assays are themselves complex biological processes that can respond to chemical perturbations in multiple ways, so determining the levels of interference of the chemicals from the data is non-trivial. In this talk I discuss a mathematical model that can be used to integrate data from a large number of assays and derive a measure of the probability that activity in the assay is due to activity in the intended pathway or due to activity in a number of assay interference pathways. The approach is employed to analyze data from a battery of 20 estrogen receptor assays in which 1800 chemicals were tested. The results are compared with expected data from a set of reference chemicals. The work was initiated at a one-week workshop titled ``Modeling Problems Related to our Environment,'' at the American Institute of Mathematics. The nature of the collaboration is itself of interest and I will devote some time to discussing that.