How do proficiency-based accountability systems affect inequality in academic achievement? This paper reconciles mixed findings in the literature by demonstrating that three factors jointly determine accountability's impact. First, by analyzing student-level data from a large urban school district, we find that when educators face accountability pressure, they focus attention on students closest to proficiency. We refer to this practice as educational triage, and show that the difficulty of the proficiency standard affects whether lower or higher performing students gain most on high-stakes tests used to evaluate schools. Less difficult proficiency standards decrease inequality in high-stakes achievement, while more difficult ones increase it. Second, we show that educators emphasize test-specific skills with students near proficiency, a practice that we refer to as instructional triage. As a result, the effects of accountability pressure differ across high and low-stakes tests; we find no effects on inequality in low-stakes reading and math tests of similar skills. Finally, we provide suggestive evidence that instructional triage is most pronounced in the lowest performing schools. We conclude by discussing how these findings shape our understanding of accountability's impacts on educational inequality.