Purpose ‐ This paper aims to discuss methods for measuring and comparing the performance of Portuguese schools, confronting metric and practice benchmarking exercises. Design/methodology/approach ‐ Data on the schools was collected within the context of a programme for school evaluation launched by the Ministry of Education in 2007. The paper first uses a non-parametric technique, data envelopment analysis, to assess the sample of schools using the data collected, taking a value-added approach. The results obtained are compared with the results obtained by the panels of evaluators within the national evaluation programme of schools. Findings ‐ The paper concludes that the performance of the schools under the metric benchmarking exercise does vary substantially. However, the results do not correlate with the judgements made by the panels of assessors regarding schools' results. This might be because assessors find it difficult to take a value-added approach and instead assess schools mostly on output measures, such as exam classifications and graduation rates, having difficulty in taking into account pupils' socio-economic background and other variables not under the control of the school. Research limitations/implications ‐ The paper focuses on the Portuguese context, and its findings might not be directly applicable to other contexts. Also, if other quantitative and qualitative methods were used, other results might have been obtained. Practical implications ‐ Although the study is focused on the Portuguese context, contributing towards a better understanding of Portuguese secondary school performance, it is believed it will be helpful to inform the debate on school evaluation, performance improvement and policy setting in other contexts. Originality/value ‐ The paper concludes by arguing for metric benchmarking exercises of the type proposed, to inform schools, evaluators and policy decision-makers, in combination with practice benchmarking exercises, which are better suited for qualitative aspects of performance.