We just published the in-depth performance review of Windows Workflow Foundation on MSDN Library. This paper should prove very useful in evaluating various options for use of WF in your software architecture. It details many different areas of WF and the performance characteristics of them. It also discusses what can be done to configure different performance settings in a WF project.
Please note that the 10308 number quoted above is for "empty" workflows. An empty workflow is not a useful scenario in and of itself. Read the article for the detailed scenario and many others. What I think is more interesting is that the WF runtime on the test hardware and for trivial activities can execute about that number of activities a second. The workflow model is treated as an activity also. So it's not adding much overhead to your .NET code and you can calculate roughly what overhead it will add.
Anyway, read the paper, comment and click the feedback at the bottom of it. Let us know what you think and use this information to make informed decisions about the use of WF for your software's business logic.