We use the repeated random assignment of external examiners to school institutes in Italy to investigate whether the effect of external monitoring on test score manipulation persists over time. We find that this effect is still present in the tests taken one year after exposure to the examiners, and is stronger for open-ended questions, for small school institutes, and for institutes located in the northern and central regions of the country. In the second year after exposure, however, this effect disappears, suggesting that monitoring is a symptomatic treatment rather than a cure of score manipulation. We discuss learning, reputational concerns, peer pressure and teacher preferences as potential mechanisms behind our findings, and present some evidence on the role played by social capital and high stakes.