A recent development in educational accountability is a risk-based approach, in which intensity and frequency of school inspections vary across schools to make educational accountability more efficient and effective by enabling inspectorates to focus on organizations at risk. Characteristics relevant in predicting which schools are "at risk on adverse effects" and robustness of results of risk-based analyses over multiple cohorts were assessed by an empirical analysis of Dutch primary schools. Adverse effects were defined as below average final achievement and/or below average value added. School composition, previous underperformance, insufficient judgments on having a systematic evaluation approach, evaluation of support, and monitoring student performance appeared as factors related to subsequent underperformance of schools. Although a rich set of possible risk factors was available, further investigation of a large number of schools is required in order to find nearly all underperforming schools. However, a group of about 40% of the schools showed very small risk on underperformance, which represents the efficiency gain when risk-based school accountability would be applied. Furthermore, whether schools are (in)accurately classified in the risk analysis as "at-risk" schools depends heavily on the chosen caesura.