Abstract Background The objective of this study was to examine whether increased levels of sitting time and physical activity in one period (within-day) or on one day (between-day) were predictive of lower levels in these behaviours in the following period or day among children. Methods Children aged 8–11 years from 8 primary schools located in Melbourne, Australia, wore an activPAL for 7 consecutive days ( n  = 235; 53 % boys). Sitting, standing and stepping time were derived for each day and for specific periods on weekdays and weekend days. Multilevel analyses were conducted using generalised linear latent and mixed models to estimate associations between temporally adjacent values (i.e. pairs of days; pairs of periods within-days) between the outcome variables. Results Significant associations were observed between temporally adjacent days and periods of the day. On any given day, an additional 10 min of stepping was associated with fewer minutes of stepping (~9 min; 95 % CI: −11.5 to −6.2 min) and standing (15 min; 95 % CI: −18.8 to −11.1 min) the following day. Greater time spent sitting during one period, regardless of being a weekday or weekend day, was associated with less time sitting and more time standing and stepping in the following period. Conclusions The direction of the results suggest that children appeared to compensate for increased sitting, standing, and stepping time both within- and between-days. The implications of such associations for the design and delivery of interventions require consideration.