According to the microboligists, changes in to DNA proceeds at a specific predictable pace. This is particualrly true on the 'junk' dna that doesn't appear to have any genetic function associated with it. by comparing the DNA of two closly related species, and applying statistics to find out how 'far' apart they are in simularity, a prediction can be made on how long ago their common ancestor existed.
So far, when a fossil gets discovered that appears to be the 'parent' specie of the two 'offspring' species (based on morphology), the age of the fossil is very close to what was predicted by the molecular biologists using the rate at which DNA will change.
There have been some cases where the 'molecular clock' thought thwere were be fossils several million years older than what was found as the assumed common ancestor, and then fossils would be found that are older, and fit much more in line with what the 'molecular clock' predicted.
I was rather skeptical of the claims for a long time when the methodology was first made public, but I will have to admit that they have made some pretty interesting predictions that have shown themselves to have merit. I am still not convinced that it is as accurate as the proponents think it is, but they have built a pretty good case for their methods.