Although there is an increasing focus on trend analyses of adolescent mental health, yet too little attention is paid to the methodological challenges and pitfalls inherent in this type of analyses. The purpose of the study is to analyse the psychometric properties of a Finnish instrument on psychosomatic problems, with a major focus on Differential Item Functioning (DIF) across time. Questionnaire data collected in 1994, 1998, 2006 and 2014 among Finnish schoolchildren in grade 9 (15-year-olds) as part of the Health Behaviour in School-aged Children (HBSC) study were utilised. The polytomous Rasch model was used to examine the psychometric properties of a composite measure of psychosomatic problems. The results clearly indicate that the composite measure on psychosomatic problems consisting of nine items does not work invariantly over time. In particular, the item depressed shows DIF across years of investigations. This item works quite differently at the first year of investigation compared to the last year showing higher expected values 2014 (=less frequent problems) than 1994. This DIF affects the person measure of change in psychosomatic problems between 1994 and 2014. Resolving the item depressed for year of investigation DIF, or removing it, increases the difference in person mean values between the two years, implying increasing psychosomatic problems over time. Since the DIF affects the trend results, different options to address the problems need to be considered. Removing the item depressed would bring the Finnish measure of psychosomatic problems in better accordance with the content of the questions on psychosomatic problems in the international HBSC protocol in which the item depressed is not included.