Considerable research has evaluated the role of accruals in determining informative earnings, with an accrual-cash flow relation at the centre of the investigation. However, much of the research is based on a misunderstanding. First, accruals are identified as the numbers that reconcile earnings to cash flows in the cash flow statement. But these are not the non-cash accruals applied in determining earnings in the accrual accounting system; rather, they are changes in balance sheet items, most of which are the relevant accruals reduced by cash flow. Thus, they are in part determined by cash flows. Second, accruals are characterized as an adjustment to cash flows, to reduce volatility of cash flows. Consequently, a negative correlation between accruals and cash flow-the accruals-cash flow relation-has been taken as the criterion for quality accruals. However, the correlation with cash flows is spurious, for the so-called accruals are determined in part by cash flows. This paper presents a corrective analysis under which non-cash accruals are identified as the components of earnings that do not involve cash flows. With this correction, the paper then conducts empirical tests that re-examine hypotheses about accruals tested in previous research, reporting contrasting results.