We are carrying out a three-year research , which purports, among other aspects, to assess to what extent self-concept of students has an influence on their academic achievement. There is some data available about this topic (Marsh, 1990; Muijs, 1997; Valle, González Cabanach, Núñez and González- Pienda, 1998; Valle, González Cabanch, Núñez, Suárez, Piñeiro and Rodríguez, 2000). The conclusions herein exposed are the result of our first year of work. We have chosen SF5 Questionnaire as measurement instrument (Self-Concept Questionnaire Form 5) of García and Musitu (2001). This questionnaire comprises 30 items divided into five sub-scales, each of them containing five items, and evaluates a dimension of academic/work, social, emotional, family and physical self-concept. Subjects should attribute a value ranging from 1 to 99 depending on their agreement with the content of each item. We have carried out a validation of the questionnaire in order to verify the validity results presented by its authors, with a 545- student sample from the two public universities of the city of Valencia (Spain): University of Valencia/Estudi General (UVEG), with approximately 46.000 students and the Polytechnic University of Valencia (UPV), with approximately 30.000 students. There were 319 students from UVEG and 226 students from UPV in the sample. Previously, we had carried out a representative sampling of students from both universities, with a level of confidence of 95% and a maximum margin of error of 5%. We found out an excellent degree of internal consistency. Cronbach's alpha coefficient was .816 for the entire questionnaire (the coefficient found by authors was .815) It was carried out a factor analysis of the principal components with oblimin rotation in order to verify the construct validity. We found five factors coinciding with those found by its authors: the first factor corresponds to academic/work self-concept (where Cronbach's alpha coefficient is .831 in comparison with .880 presented by the authors), the second factor corresponds to emotional self-concept (where Cronbach's alpha coefficient is .475 in comparison with .731 presented by the authors), the third factor corresponds to family self-concept (where Cronbach's alpha coefficient is .596, in comparison with .769 presented by the authors), the fourth factor corresponds to physical self-concept (where Cronbach's alpha coefficient is .742 in comparison with .744 presented by the authors), and finally, the fifth factor corresponds to social self- concept (where Cronbach's alpha coefficient is .729, in comparison with .698 presented by the authors). Therefore, the factor analysis confirms the theoretical structure supported by García and Musitu (2001).Later, it was established a correlation between the scores corresponding to the five factors and the grades obtained by students in seven subjects, through Pearson correlation, where significant positive correlations were found between grades and the first (academic/work self-concept), third (family self-concept) and fourth (physical self-concept) factor. However, this correlation was not found between grades and the second (emotional self-concept) and fifth (social self- concept) factor.Also, it was carried out a cluster analysis following k-means procedure presenting factor scores as variables in order to establish student groups formed depending on their self-concept. Three groups were found. The first group, consisting of 76 students, with a low self-concept in every dimension or sub-scale; the second group, consisting of 248 students, with a high academic and family self-concept, but a rather low emotional, physical and social self-concept; and a third group, consisting of 154 students, with a high physical and social self-concept, but with a rather low academic, emotional and family self- concept. Later, the existing differences (ANOVA) in grades between the three groups were examined, and no significant diferences were found in any case. However, mean scores of grades were higher in subjects from the second and third group, with a better self-concept, in comparison with students from the first group with a low self-concept. The results would seem to confirm the existing relation between self-concept and achievement (González, Mendiri and Arias, 2002; Valle, González Cabanach, Núñez and González-Pienda, 1998; Valle, González Cabanach, Núñez, Suárez, Piñeiro and Rodríguez, 2000), however, they are not conclusive and we intend to continue examining these results with new and more complex analysis.