Prior studies provide mixed evidence about whether the accrual anomaly exists in the Korean stock market. We seek to reconcile the mixed evidence by applying a different measure of accruals—accruals scaled by earnings (percent accruals)—in comparison to accruals scaled by total assets (traditional accruals). Based on 9399 firm–year observations for the 1994–2010 period, we find evidence of the accrual anomaly in the Korean stock market when using percent accruals but not when using traditional accruals. Of particular note is that when firms are sorted by traditional accruals, the lowest accrual decile includes firms with low cash flows, which leads to low returns that eliminate the abnormal returns of the accruals-based trading strategy. This also occurs when we use other firm-size proxies to deflate accruals. In contrast, when using percent accruals we find consistent evidence that the accrual anomaly exists regardless of research design specifications or sample selection criteria. Our findings suggest that percent accruals are a useful alternative to traditional accruals, especially in markets where the lowest traditional accrual decile exhibits very low returns.