We analyze prestack PS migration images and their focusing sensitivity to errors in the computed PS traveltimes. The key analysis tool is a formula that defines PS traveltimes errors as explicit functions of velocity model errors. The most important factors in this formula are the PS velocity and the P-to-S velocity ratio. Analysis shows that the error in PS traveltime for shallow events is usually larger than that for deep events for a given error in the velocity model. Also the PS traveltime is affected more severely by errors in the PS-velocity model than in the P-to-S velocity ratio. The effect of traveltime errors increases with dip angle of reflectors. Numerical analysis shows that, for a fixed scatterpoint, the effect of the PS-wave velocity error is several times larger than the effect of the error in the P-to-S velocity ratio. Examples from field data show that the PS-wave velocity must be estimated accurately with errors less than 1% in order perfectly flatten the events in common-image-point (CIP) gathers. In contrast, an error in the PS-velocity ratio of up to several percent is allowed. This suggests that for acceptable PS-wave migration, only the PS-velocity model and a rough estimate of the P-to-S velocity ratio is needed. This finding is useful for processing PS-wave data because it is difficult and time consuming to estimate the velocity ratio accurately from the real data. This finding is also useful for our understanding of PS-wave behavior and for PS-wave imaging.