This study tests the bonus-maximization hypothesis that managers make discretionary accrual decisions to maximize their short-term bonuses. By using the management and financial reporting database of a large conglomerate, we extend previous investigations in two ways. First, the analysis is conducted using business unit-level data, which reduces the aggregation problem that is likely to arise using firm-level data. Second, managers in this setting are paid bonuses based solely on business unit earnings. The potentially confounding effects of long-term performance and stock-based incentive compensation are thus absent. These innovations yield robust evidence consistent with Healy (1985).