Many meta-analytic reviews have examined the cognitive benefits of regular physical exercise across the lifespan, with a wide range of inclusion criteria and disparate analytic approaches. In fact, most global health organizations incorporate the purported cognitive benefits as part of their policies to promote regular physical exercise. We assess whether those claims are well supported by scientific evidence. We conducted a pre-registered (PROSPERO: CRD42020191357) umbrella review of meta-analyses limited to randomized controlled trials (RCTs) on exercise-cognition. Outcomes of each meta-analysis were reanalyzed following a multilevel model approach. Publication bias was also analyzed to correct final effect sizes. Most of the 12 reviewed meta-analyses reported a positive small effect (d = 0.21, [0.07 0.40]) of physical exercise across all cognitive domains in all life stages. However, a critical assessment of the primary studies included in these reviews and the meta-analytic approaches used indicates that the evidence might not be reliable and therefore no strong conclusions should be drawn based on the existing evidence. We found that the literature has undergone an unusually rapid growth upon underpowered RCTs and potentially biased meta-analytic reviews. The exponential accumulation of low-quality evidence on the exercise-cognition topic has led to stagnation in the field, hindering the discernment of the potential impact of exercise on cognition. We urge worldwide organizations committed to public health to reconsider their recommendations on the promotion of regular physical exercise to boost cognitive functions in the healthy population until high-quality empirical evidence confirms these exercise-induced cognitive benefits.