This paper investigates the validity of a national, standards-based teacher evaluation programme by examining the relationship between teachers' evaluation results and their students' learning progress. We used census achievement data that assessed the same cohort of students at the end of 8th and 10th grade. We applied multilevel modelling and calculated correlations between teachers' value-added estimates and their results based on the evaluation system's multiple measures of teaching quality, which include a portfolio of classroom teaching artefacts and a video-taped lesson, a peer interview, supervisor assessments and a self-assessment. Our study finds a medium-size positive relationship between value-added estimates in mathematics and teachers' portfolio results, while relationships with the other instruments are weaker. Correlations are lower for Spanish, but showing a similar pattern across instruments. We also find significant yet weak relationships between value-added estimates and three of six emerging factors of teaching practice based on the evaluation's portfolio data.