Model checking tools have been effective in testing concurrent software represented by communicating finite-state machines. But these tools may require a very large amount of memory. A finite-state model can be translated automatically into a compact AND-OR graph. We use an abductive random search scheme to extract, from the AND-OR graph, information about the execution of the program represented by the original finite-state model. We use the search to measure testability. For AND-OR graphs representing highly testable programs, we find quickly everything it is possible to find; that is, if the number of unique goals found is plotted, we see a quick rise to a level plateau. The search can also be used to prove simple logical properties. To determine what makes a finite-state model more or less testable, we analyze random search results for 15,000 randomly generated models with a range of attributes. We also show how this technique can be used on a model much too large for model checking tools. v 4.9 Dekker's solution to the two-process mutual exclusion problem, as Promela from [10] (left) and as finite-state machines output by SPIN (right).. .. .. 4.10 Promela model from Figure 4.9 (originally from [10]) as finite-state machine