Background It is presumed that physical activity (PA) level declines during the lifespan, particularly in adolescence. However, currently, there is no study that quantifies these changes and pools results for a common interpretation. Therefore, the purpose was to systematically review the international literature regarding PA change during adolescence, and to quantify that change according to a series of study variables, exploring gender-and-age differences.

Methods An electronic search was conducted in the Medline/PubMed and Web of Science databases. Longitudinal studies with, at least, two PA measures throughout adolescence (10–19 years old) or the first PA measure during childhood and the second one during adolescence were selected. From each article, study project name, country, year of the first data collection, sample size, baseline age, follow-up duration, characteristics of the instrument (type, recall time, PA intensity and PA domain), unit of PA measure and report of statistical significance were collected.

Results Overall, 26 studies matched the inclusion criteria. Most were carried out in the USA, assessed PA by questionnaire, and found a decline in PA levels during the adolescence. On average, the mean percentage PA change per year, across all studies, was −7.0 (95% confidence interval: −8.8 to −5.2), ranging from −18.8 to 7.8. The decline was significant according to most sub-groups of variables analysed. Although earlier studies revealed a higher PA decline in boys, the decline has been greater in girls in more recent studies (commenced after 1997). Moreover, although the decline among girls was higher in younger ages at baseline (9–12 years), it was higher in older ages (13–16 years) among boys.

Conclusions The decline of PA during adolescence is a consistent finding in the literature. Differences between boys and girls were observed and should be explored in future studies. Interventions that attempt to attenuate the PA decline, even without an increase in PA levels, could be considered as effective.