Is There a Domain Wall Problem?

We show that spontaneous breaking of discrete or continuous symmetries does not necessarily imply either symmetry restoration or the thermal production of defects at high temperature (at least up to $T\ensuremath{\sim}{M}_{\mathrm{Planck}}$). This may imply that there is no domain wall problem. As an example we show how this applies to the Peccei-Quinn scenario.