Methods and apparatus sort integrated circuits by maximum operating speed (f.sub.max). The timing for a first set of critical timing paths is statistically characterized. The first set can be, for example, the set of all critical timing paths. For example, the timing can be generated by using static timing analysis (STA). The timing for a second set of critical timing paths is statistically characterized. The second set can be, for example, a sample set of critical timing paths that are measurable or are measured for a device during test. The timing can be based on STA, derived from a known good device, and the like. A device under test (DUT) is tested, and the timing for the second set of critical timing paths is determined. A fitting technique is used to fit the expected device characteristics and the measured data for the DUT, and in one embodiment, the parameters used for fitting are applied to the first set of critical timing paths, which are then used to determine an appropriate f.sub.max for the DUT.