Zero the Z position indicator and slice again. You should find that the surface of the sample remains at around the same depth (+/- about 5 microns). If the depth varies by more than about 10 microns, cut again and re-measure. Note that the surface may not be flat in some regions. e.g. if you're cutting from the cerebellum to the bulb, then you should reference from the pons as this area is physically more stable than the cerebellar hemispheres (at least around the transition region between the cerebellum and cortex). Usually the surface height stabilises within a 3 cuts. Sometimes it takes longer and rarely it alternates in a way that suggests the system is alternating between thick and thin sections. If this happens, switch to cutting thicker sections. If this is not an option, try cutting a couple of slightly thicker sections then switch back to your intended section thickness. Once the sample surface is stable, move the objective down to about 20 to 35 microns beneath the surface. This will be your first imaging depth.