Resumen: In this work, we present some results obtained from the analysis of the behavior of 91 mathematics teachers (prospective, secondary education and university) when they grade three different types of correct answers to a classical high school problem through a questionnaire. In addition to a descriptive analysis that studies the variability and the interrater reliability, we analyze the role of experience and training as well as the influence of the different solving methods. Furthermore, we try to identify profiles of correctors among secondary education teachers using both quantitative (cluster analysis) and qualitative (content analysis) methods. In particular, we observe a great variability on the assigned grades as well as a low interrater reliability. The belonging to a particular group has impact over the assigned rates while experience has no significant influence. The grades are higher when methods closer to the corrector are used. Finally, we have been able to identify three different clusters, which are determined by the comments and actions regarding three aspects of the students’ answers: argumentation, correctness and method. Idioma: Inglés DOI: 10.4438/1988-592X-RE-2015-371-307 Año: 2016 Publicado en: Revista de educación 2016, 371 (2016), 31-55 ISSN: 0034-8082 Factor impacto JCR: 1.185 (2016) Categ. JCR: EDUCATION & EDUCATIONAL RESEARCH rank: 109 / 235 = 0.464 (2016) - Q2 - T2 Factor impacto SCIMAGO: 0.68 - Education (Q1)