A catch activity can include a fault name and/or a fault variable. The fault variable can be defined as a WSDL message type or a schema element. The schema element can be a member of a substitution group defined in a WSDL's schema, or it can be a single-part message defined as an element.

The fault matching logic for WS-BPEL is a best match strategy. This means that the available catch elements are analyzed to determine which catch is the best match for the given fault. Contrast this with the fault matching logic for a language like Java which has a first match strategy. Because of the best match strategy, the order of the elements within the fault handlers does not affect which catch element is matched.

It is possible that more than one catch will match to a given fault. In this case, the catch that is matched with the higher priority rule is selected. In the case where two catch elements are selected by the same rule, then the best match is determined by whichever catch element declares an element variable that is closer to the fault element in terms of its substitution group hierarchy.

The priority system first considers whether a catch has a only a fault name or a fault name with data. The highest priority is for a catch with only a fault name and no data.

For catches with fault data, Process Developer executes catches, in the order described in the following table.

For details on execution rules for catch/catch all, see Fault Handling Processing Rules.