Does Measuring Code Change Improve Fault Prediction?
Code churn measures seem to be very nice predictors of software failures. In this paper, Bell, Ostrand and Weyuker show that one does not need much more. They also investigate whether code churning is often linked to other anomalies based on 18 releases of a large software system.
I recommend this paper for two reasons - it shows how simple measures can give a lot of value and because it comes from top names!