A fault location
estimation system comprises single-fault-assumed diagnostic means that assumes a
single fault and stores fault candidates, fault types, and detected error-observation nodes at which an error arrives from the fault candidates; error-observation node basis candidate classification means that classifies error propagating fault candidates into groups according to error-observation nodes using the fault candidates and the error-observation nodes and stores the groups as fault candidate groups; inclusion fault candidate
group selection means that acquires a relation between each fault candidate group and a fault output, calculates an
inclusion relation among the fault candidate groups, and, if path information on one fault candidate group includes path information on another fault candidate group, deletes the inclusion fault candidate group; inter-pattern overlapping means that calculates combinations of fault candidate groups that can reproduce a test result in all test patterns by referencing the fault candidates and the fault candidate
classification result and extracts fault candidates that are common to fault candidate groups, calculated in multiple patterns, to create a new fault candidate group; and multiple-fault
simulation checking means that selects fault candidates from the combinations of fault candidate groups, one from each fault candidate group, performs multiple-fault
simulation, compares the
simulation result with the test result, and outputs a combination of fault candidate groups which matches relatively well with the test result and has a relatively high fault possibility.