We undertook an exclusive meta‐analysis of cohort studies investigating the possible link between prenatal selective serotonin reuptake inhibitor (SSRI) exposure and autism spectrum disorders (ASD) in children to further investigate our previous suggestion of confounding by indication. The point estimates regarding the following cohorts were extracted and pooled: (1) pregnant women who discontinued SSRI until 3 months before pregnancy; (2) pregnant women who were exposed to SSRI during pregnancy; and (3) pregnant women with maternal psychiatric disorder but no exposure to SSRI during pregnancy. Although the pooled point estimate of the first cohort showed a trend for increase, it did not reach significance. The pooled point estimates of the latter cohorts showed a significant association with ASD which strengthens our previous suggestion of confounding by indication. Future studies should be adequately designed to differentiate whether the previously suggested association is a result of maternal psychiatric disorder or SSRI exposure or both.