Propensity scores represent an intriguing alternative method to reduce the impact of confounding variables, especially when there are multiple potential confounding variables. This paper considers a range of models comparing the propensity score approach to the more traditional approaches of adjusting for confounders. I think the conclusions are overly simplistic, but the paper is still worth reading.
M. Soledad Cepeda, Ray Boston, John T. Farrar, Brian L. Strom. Comparison of Logistic Regression versus Propensity Score When the Number of Events Is Low and There Are Multiple Confounders. American Journal of Epidemiology. 2003;158(3):280–287. doi:10.1093/aje/kwg115.
You can find an earlier version of this page on my blog.