Post-interventional successes apart, our data did not support maintained intervention benefits over time. This is in contrast with two similar studies with data on long-term follow-up on biological risk factors, which do provide tentative support of positive sustainable benefits. These studies were based on provision of additional physical education and have evaluated both immediate post-intervention and long-term follow-up difference between intervention and control schools [
41,
42,
51]. The CoSCIS study was a non-randomized controlled intervention based on a doubling of weekly physical education from 90 to 180 min for 3 years in addition to provision of outdoor play-ground equipment. The CoSCIS trial observed higher insulin sensitivity in intervention school boys post-intervention and an approximately 3 mmHg lower systolic blood pressure change in boys when followed-up 4 years after the intervention had ceased [
42]. No differences between intervention and control school girls were observed at any time point. At the 4-year follow-up of the KISS randomized controlled trial which included nine months of augmented (lesson content) and expanded (from 135 to 225 min weekly) physical education curriculum at its core, participants at intervention schools had a substantially higher cardiorespiratory fitness (5% difference) and greater participation in leisure-time sports compared to participants at control schools [
51]. At follow-up, only one risk factor was affected, which was in contrast to the post-intervention evaluation where several metabolic markers (composite score, HDL-cholesterol, triglyceride, glucose), cardiorespiratory fitness, and body composition were favourably modified by the intervention [
41]. Long-term differences between intervention and control schools are particularly interesting in the light of the discontinued interventional support in both of these studies. However, even though these sustainable differences would have public health relevance if further maintained over time, statistical significance of results amounts to only one of eleven [
42] and one of fourteen [
51] investigated outcomes. Thus, in conjunction with high rates of attrition (439 of initially 694 participants [
42] and 293 of initially 502 participants [
41] available for long-term follow-up, (these losses apart from post-randomization but pre-baseline drop-out)), a cautious interpretation of cause and effect is warranted. Intervention “dose” does not appear to explain the apparent lack of sustained benefits in the CHAMPS-study DK in comparison with the CoSCIS and KISS studies. The CHAMPS-study DK provided an additional 3 h of physical education per week in the intervention group which was double the dose of additional physical educations as delivered in the CoSCIS study. The KISS study was based on daily 45 min of augmented physical education but in comparison with the control condition, only an additional 90 min per week was added. As effects of physical education on physical activity levels do not appear to extend beyond the days when physical education is taking place [
18] this could suggest intervention content should be delivered daily. Accordingly, the Sogndal Study was based on 60 daily minutes of physical activity and achieved remarkable effects on biological risk factors [
40]. However, a fairly small Icelandic study providing 60 min of daily physical activity in the intervention group did not observe any effect on cardiovascular risk markers after 2 years of intervention [
48]. The Cretan Health and Nutrition Education Programme provide further support that sustainable benefits of quality physical activity in school can be achieved. Following 6 years of a comprehensive school health programme (diet, physical activity, and risk behaviours), in which augmented physical education (two lessons per week) was one component, differences between intervention and control schools in blood cholesterol and blood pressure were maintained 4 years post-intervention [
52,
53]. Selection bias (attrition and non-randomization) appears the major threat to the validity of the long-term follow-up of the Cretan Health and Nutrition Education Programme. Noticeably, control schools in the Cretan study did not receive any structured physical education for the first half the intervention period which may limit generalizability of the study to a large number of school systems. Conversely, an Australian twenty-week randomized controlled trial including daily twenty minutes fitness sessions did not results in immediate or six-months post-intervention effects on blood pressure or cholesterol levels in high- or low risk children [
54]. The internal validity of the Australian study appears robust to selection bias owing to randomization and low attrition. However, the observed lack of short-term effectiveness is potentially explained by the apparently non-trivial time-frame from cessation of the intervention to collection of post-intervention data. Also, the change in BMI appeared larger in intervention group boys as compared to boys in the control condition.