Exposure to prenatal maternal stress (PNMS) has been implicated as a risk factor for a range of psychiatric disorders in children. However, there have been a few studies showing inconsistent associations between PNMS and offspring autistic-like behaviours. We therefore aimed to examine whether trimester-specific PNMS exposure might be related to an increased risk of autistic-like behaviours among preschoolers. Using data from Longhua Children Cohort Study, mothers of 65,931 preschool children were asked to recall their level of PNMS in each of the three trimesters of pregnancy, while children's current autistic-like behaviours were assessed using the Autism Behaviour Checklist. A series of Cox regression models were fitted to assess the association between PNMS exposure and autistic-like behaviours. After adjusting for potential confounders, the Cox regression models showed that PNMS exposure, especially during the second pregnant trimester, was significantly and positively associated with the presence of children's autistic-like behaviours. The strength of these associations was enhanced with the increase of PNMS exposure level. Furthermore, based on different permutations of exposure versus no exposure in each trimester, the participants were divided into eight groups. A cross-over analysis confirmed the aforementioned finding that the second pregnant trimester might be the sensitive period for PNMS exposure increasing the risk of autistic-like behaviours. Our findings supported the hypothesis of an association between PNMS exposure and autistic-like behaviours among preschoolers. Preventive interventions should be trialled to examine whether minimizing maternal psychological stress during pregnancy, especially the second trimester, may reduce the risk of offspring autistic-like behaviours. © 2020 John Wiley & Sons Ltd.