Only certain galaxies are included in surveys: those bright and large enough to be detectable as extended sources. Because gravitational lensing can make galaxies appear both brighter and larger, the presence of foreground inhomogeneities can scatter galaxies across not only magnitude cuts but also size cuts, changing the statistical properties of the resulting catalog. Here we explore this size bias and how it combines with magnification bias to affect galaxy statistics. We demonstrate that photometric galaxy samples from current and upcoming surveys can be even more affected by size bias than by magnification bias.