Matt Chingos and Guido Schwerdt have posted a Harvard Program on Education Policy and Governance working paper with the first rigorous analysis of the effects of virtual education in K-12. You can read it online as well as look at this excellent summary by Marty West on the Ed Next blog.
The bottom line is that Florida students taking Algebra and English I online tended to do at least as well as those who took those courses in traditional classrooms, controlling for prior achievement and demographic characteristics. To strengthen the causal identification the authors focus on comparing students who took at least one online course so they would be more alike in the unobserved characteristics that might motivate a student to take courses online.
Faring equally well is a positive outcome for online education because delivering education virtually has the extra benefits of expanding access to students at schools that do not offer those classes. Delivering courses online is also considerably cheaper.
Of course, this is one study and we are in the early days of developing virtual education, so these findings may not apply to future circumstances. But they are certainly encouraging enough to continue expanding virtual education and collecting information on the results.