Introduction Physical activity (PA) is one of the most important lifestyle behaviors to reduce the risk for non-communicable diseases. Positive effects of the urban environment on PA in terms of walkability measures such as street connectivity, land use mix, and availability of public open spaces in the urban neighborhood are well investigated. However, evidence of this association is mostly based on cross-sectional studies. Usually, discrete measures of urban point characteristics such as the simple intensity are calculated based on arbitrary municipal districts inducing scaling bias. The present study aims to longitudinally investigate the effect of built environment characteristics on PA of children in their transition phase to adolescence. We considered a kernel intensity approach to assess built environment characteristics and linked urban measures with objectively measured PA data that were collected in the IDEFICS/I.Family cohort. Methods Spatial data were collected in six study regions of three countries, Germany, Italy, and Sweden, participating in the IDEFICS/I.Family cohort. We considered walkability measures such as population density, land use mix (LUM), street connectivity, and availability of public transit. We additionally assessed the availability of public open spaces (POS). Point characteristics, e.g. intersections or POS, were assessed via adaptive anisotropic kernel functions. Urban measures were calculated in network-dependent home neighborhoods using a network distance of 1 km. We used standardized z-scores of urban measures to construct a moveability index that quantifies urban opportunities for PA of children and adolescents. Based on the place of residence, we linked urban measures with individual-level data of 699 children aged 2 to 14, who provided at least two (n = 610) or three (n = 89) accelerometer measurements, from either baseline, first follow-up (IDEFICS study), or second follow-up (I.Family study) with in total 1487 observations. Trajectories of moderate-to-vigorous physical activity (MVPA) as well as light physical activity (LPA) were modelled using linear mixed models accounting for repeated measurements nested within individuals by means of a random intercept and random linear slope for age (centred at 8 years). Environmental variables (z-scores) were included as fixed effects and as interaction effect with age. All models were stratified by sex and adjusted for age, BMI z-score, parental education, valid wear time, season, and region. Results Mean age was 6.6 in boys and 6.8 in girls. Compared to boys, girls spent on average less minutes in MVPA (boys: 58.8, girls: 51.6) and LPA (boys: 305, girls: 294) per day. Linear trajectories showed a significant decline of MVPA over age (boys: = −2.9, 95%CI: [−3.5; − 2.3], girls: = −3.6, [4.1; − 3.0]) and LPA (boys: = −7.6, [−9.1; − 6.1], girls: = −7.6, [−9.1; − 6.1]). The moveability index interacting with age showed only small effects on PA trajectories that were more pronounced for MVPA in boys (= 0.08, [−0.08;0.24]) and for LPA in girls (= 0.15, [−0.23;0.50]). Considering POS a stronger interaction with age was found for MVPA in boys (= 0.24, [−0.30; 0.78]) than in girls (= −0.08, [−0.58; 0.42]). The interaction of population density with age showed a positive effect on MVPA in boys, but not in girls, while the effect on LPA was stronger in girls compared to boys. Only in Girls, negative effects were found for LUM on MVPA (= −0.22, [−0.76; 0.33]) and LPA (= −1.14, [−2.52; 0.23]). Conclusion The kernel approach allowed for a flexible modelling of urban measures and improved the assessment compared to simple density methods. Although PA strongly declined with age, urban measures showed a supportive effect that attenuated this decline in the transition phase from childhood to adolescence. However, urban measures differently affected PA in boys and girls.