A method and apparatus for reducing the effects of non-repeatable runout (NRRO) in the estimation of repeatable runout (RRO) in a disc drive is disclosed. PES information is a acquired in a way that leads to randomization of the phase of the NRRO, thereby decreasing the number of revolutions required to estimate the true RRO. A number of RRO estimation measurements are taken over a number of non-consecutive disc revolutions that are spaced in time. In one embodiment, one revolution of RRO data is collected from each head before collecting a second revolution of RRO data from any of the heads. In another embodiment, RRO estimation is concurrent with media certification, such that PES data is obtained while the head is in the write position and writes a track of data for media certification. Similarly, PES data for the read position is obtained while concurrently reading back the written data.