We study how securities analysts influence managers’ use of different types of earnings management. To isolate causality, we employ a quasi-experiment that exploits exogenous reductions in analyst following resulting from brokerage house mergers. We find that managers respond to the coverage loss by decreasing real earnings management while increasing accrual manipulation. These effects are significantly stronger among firms with less coverage and for firms close to the zero-earnings threshold. Our causal evidence suggests that managers use real earnings management to enhance short-term performance in response to analyst pressure, effects that are not uncovered when focusing solely on accrual-based methods.