Objective: To test Goodman's theoretical 5-factor model of the Strengths and Difficulties Questionnaire (SDQ) in an analysis of the Norwegian parent (P) and teacher (T) versions of the questionnaire. Method: The T-SDQ was analyzed for 8,999 (95.4% of all) children in primary school grades 2-4 in Bergen, Norway, whereas the P-SDQ was examined for 6,430 children (68.2%). Main analyses were exploratory and confirmatory factor analyses. Results: The analyses supported a modestly modified version of Goodman's 5-factor model for both the P-SDQ and the T-SDQ. The correlations between latent factors were high, particularly when the conduct problem factor was involved, reflecting a high level of overlap between the problem areas. Cross-informant correlation between parents and teachers was moderate to high for the problem subscales and considerably higher than the corresponding correlation between subscales that has been found in earlier studies. Conclusions: None of the alternative models derived from the exploratory factor analysis fitted the data better than a slightly modified version of Goodman's 5-factor model, which showed acceptable goodness of fit. (Contains 8 tables and 1 figure.)