ABSTRACT: We document how the effectiveness of an accruals-based trading strategy changes with the benchmark used to identify an extreme accrual. We measure “percent accruals” as accruals scaled by earnings, rather than total assets, and show that this seemingly small change produces a radically different sort of the data. We find that a trading strategy based on percent accruals yields significantly larger annual hedge returns than the traditional accruals measure, and does so mostly by improving the long position in low-accrual stocks. The hedge returns are also significant in all but the lowest quintile of arbitrage risk. We show that percent accruals more effectively select firms where the difference between sophisticated and nai¨ve forecasts are the most extreme. As such, our results are consistent with the earnings fixation hypothesis and are inconsistent with some alternative explanations for the accrual anomaly.