It is common practice in process synthesis to choose a reactor structure and use the free parameters associated with it to optimize the system. It is of some importance to assess how effective such an approach is. A geometric (attainable region) approach is used to find the bounds on the performance of reactor systems, and these are compared with the complete attainable region. It is shown for instance that the conversions obtained in segregated and maximimum mixed reactors are not general bounds on conversion. Even two or more environment reactor models need not be sufficiently general to provide bounds on conversion. Given a proposed reactor system, it is thus possible to compare the achievable conversions with those of the complete attainable region. This provides a useful test to determine whether significant improvements in reactor performance can be obtained by using alternative reactor structures.