Crucial to our understanding of psychopathology in children is information on the stability and change of behavioral/emotional problems and competencies across time. Existing studies, reviewed in this article, provide only limited answers due to one or more of the following factors: restriction of the sample to a single locality, the use of selected samples, large sample attrition, the use of different assessment instruments at different times, the small number of specific behavioral/emotional problems or the use of very broad categories of functioning. The present study was designed to test the persistence and change of behavioral/emotional problems and competencies as reported by parents of 1,412 children aged 4-14 over a two-year period in a representative sample of the general population. At both points, standardized parents' reports were obtained using the Achenbach Child Behavior Checklist (CBCL). A recovery rate of 80.2% was obtained. Our study is the only one, as far as we known, to present follow-up data separately for ages covering the preschool period to adolescence, making it possible to compare stabilities in the occurence of behavioral/emotional problems across various developmental levels. Over the two-year follow-up period, there was a significant decrease in problem scores. Although possible, it is unlikely that the mental health of children in our sample increased very much. A decrease in problem scores over time has been demonstrated for clinical as well as general population samples in other studies as well. No satisfactory explanations are available for this phenomenon. It is important, however, that studies employing a retest design take account of this phenomenon. Irrespective of changes in the magnitude of group scores over time, stability coefficients can tell us whether individual children tend to preserve their rank orders. The stability coefficients between Time 1 and Time 2 total problem scores ranged from .62 (girls, aged 4-5) to .71 (boys, aged 6-11). No significant differences between stability coefficients for different age-groups and for both sexes could be demonstrated. The changes in scores across time for individual items were consistent with the cross-sectional data of our earlier study (Verhulst et al. Acta Psychiatrica Scandinavica, 1985, 72, suppl. 323). Stability coefficients for social competence scores at both times of assessment were generally somewhat weaker than those found for problem scores.(ABSTRACT TRUNCATED AT 400 WORDS)