Functional coverage models which measure the sufficiency of test stimuli are essential to the verification process. A key source of difficulty in their deployment emanates from the manual and imprecise nature of their development process and the lack of a sound measure of their quality. A functional coverage model can be considered complete only if it accurately reflects the behavior of the Design under Verification (DUV) as described in the specification. We present a method to automatically generate coverage models from a formal CTL description of design properties. Experimental results show that the functional coverage models generated using our technique correlate well with the detection of randomly injected errors into a design.