We investigate whether and how physicians' prescriptions of a new drug are influenced by their colleagues in the same hospital during shared working time. We use longitudinal data of physicians who prescribed antipsychotic drugs for schizophrenia patients in Taiwan between 1997 and 2010. We find that peer effects are small, but stronger among physicians of similar age and among those sharing a longer, larger, or more stable group. Peer effects are also stronger when drugs are newly introduced. We also find that peer effects are more likely to be overestimated using fixed-effect models than using first-difference models. (JEL D01, D83, I10)