In the present study in preschools, average step count values per minute were 65 in boys and 54 in girls. The higher activity levels in boys are in line with the literature [6, 7, 13, 14, 17, 28] and builds on to the evidence that lower levels of PA in girls compared to boys are already significant at young age.
The step count values of the present study were considerably higher than the average step counts reported by Boldemann et al , in 4- to 6 year olds, during preschool attendance. In the latter study boys took on average 21 steps per minute and girls took 18 steps per minute. But these data were not specific for recess time and also included structured sedentary activities, like having lunch. The present average step count values approximate taking only one step per second. Apparently during recess the engagement in vigorous physical activity was limited and possibly, as found by Mc Kenzie et al , large parts of recess times may have been spent sedentary. Hence there are opportunities to increase activity levels at recess in preschoolers. Furthermore it was found in the present study in boys and in girls respectively, that 27 % to 35 % of variance in step counts may be attributed to differences between schools.
In both genders more space per child was found to be associated with more physical activity during recess. Therefore preschools should be encouraged to provide sufficient space for recesses, if necessary by splitting into groups with different recess times. Furthermore it was observed that boys and girls took fewer steps per minute when recesses lasted longer. This may be due to the possibility that children show a burst of activity when they first go outside which subsides with time. In the present study recess times varied from 9 minutes to 50 minutes. Possibly physical activity levels decreased after a certain amount of time, due to fatigue or getting bored. Also McKenzie et al  found in preschoolers that activity levels declined as recess time elapsed. Efforts to increase children's outdoor play time need to be advocated, since children are presumably still more active overall if they are outdoors for longer periods of time. However, efforts to promote continued activity during outdoor play may be needed or more recess periods per day may be preferable.
An interesting finding of the present study is the fact that children were less active when more teachers were supervising. However this was only significant in girls. This can be explained by the fact that many teachers supervise sitting down or standing still. Since many children, and presumably especially girls, prefer to stay close to the teachers, more supervising teachers may cause decreased activity levels. Consequently efforts seem useful to inform and encourage present and future preschool teachers to promote activity during recess (e.g. by playing with the children or at least encourage active play). Incorporating physical activity promotion in the training of future preschool teachers may enable them to implement the principles in their daily work and to enter into a professional career with a positive attitude toward physical activity promotion. According to the findings of Boldemann et al  in environments with trees, shrubbery and broken ground step counts/min were higher than in delimited environments with little vegetation. In the present study height differences and vegetation were present in about half of the schools but the presence was not significantly associated with higher step counts. This can be explained by the fact that the extend of the height differences and vegetation on the playgrounds was only limited. On the other hand a harder ground surface was a borderline significant predictor for higher step counts in boys only. A possible explanation is that the spontaneous behaviour may differ between both sexes, with boys being more triggered by harder ground surfaces, which are mainly used for more sports-related, competitive activities.
A remarkable finding of the present study is the fact that the availability of toys, the presence of aiming or playing equipment, like swings or slides, and the presence of markings was not associated with more physical activity. Possibly the choice of toys (e.g. hoops), equipment pieces (e.g. swing) or markings (mainly field markings) were not optimal in the observed preschools and results may be different when focusing on certain types of toys, equipment or markings. Another explanation may be that toys and equipment often lead to standing in line to use the piece of toy or equipment. In the study of Zask et al  equipment availability was also not significant PA predictor in elementary school children, except for balls. Further study is needed to evaluate if specific toys, equipments or markings may be more successful to trigger physical activity and to evaluate if triggering may appear when they are available for all children.
A first limitation of the present study is that all data were collected during winter. Therefore physical activity levels possibly suffered from seasonal influence. However Belgium has a mild climate, measurements were only taken when the weather permitted outdoor playing and according to the findings of Fisher et al  seasonality plays only a limited role in physical activity in young children.
A second limitation is the use of pedometers, which may not capture or underestimate some activities among young children, like swinging or crawling. Strengths of the present study are the relatively large sample size, the use of an objective physical activity measure and observation of the playground environment, and the use of multilevel analyses to take into account adjustment for clustering of subjects within preschools.