Worldwide, a large number of infants, toddlers, and preschoolers are enroled in formal non‐parental early childhood education or care (ECEC). Theoretically, lower adult/child ratios (fewer children per adult) and smaller group sizes are hypothesised to be associated with positive child outcomes in ECEC. A lower adult/child ratio and a smaller group size may increase both the extent and quality of adult/child interactions during the day. The objective of this review is to synthesise data from studies to assess the impact of adult/child ratio and group size in ECEC on measures of process characteristics of quality of care and on child outcomes. Relevant studies were identified through electronic searches of bibliographic databases, governmental and grey literature repositories, Internet search engines, hand search of specific targeted journals, citation tracking and contact to experts. The primary searches were carried out up to September 2020. Additional searches were carried out in February 2022. The intervention was changes to adult/child ratio and group size in ECEC with children aged 0–5 years old. All study designs that used a well‐defined control group were eligible for inclusion. The total number of potential relevant studies constituted 14,060 hits. A total of 31 studies met the inclusion criteria and were critically appraised by the review authors. The 31 studies analysed 26 different populations. Only 12 studies analysing 8 different populations (N = 4300) could be used in the data synthesis. Included studies were published between 1968 and 2019, and the average publication year was 1992. We used random‐effects meta‐analysis, applying both robust‐variance estimation and restricted maximum likelihood procedures to synthesise effect sizes. We conducted separate analyses for process quality measures and language and literacy measures. The meta‐analysis using measures of process quality as the outcome included 84 effect sizes, 5 studies, and 6256 observations. The weighted average effect size was positive but not statistically significant (effect size [ES] = 0.10, 95% confidence interval [CI] = [−0.07, 0.27]) using robust‐variance estimation. The adjusted degrees of freedom were below 4 (df = 1.5), meaning that the results were unreliable. Similarly, the low number of studies made the estimation of heterogeneity statistics difficult. The I2 and τ2 estimates were both 0, and the Q‐statistic 2.3 (p = 0.69). We found a similar, but statistically significant, weighted average effect size using a restricted maximum likelihood procedure (ES = 0.10, 95% CI = [0.004, 0.20]), and similar low levels of heterogeneity (Q = 0.7, I2 = 0%, τ2 = 0). The meta‐analysis of language and literacy outcomes is based on three studies exploring different changes to group size and/or adult/child ratio in ECEC. The meta‐analysis of language and literacy measures included 12 effect sizes, 3 studies, and 14,625 observations. The weighted average effect size was negative but not statistically significant (ES = −0.04, 95% CI = [−0.61, 0.53]) using the robust variance estimation procedure. The adjusted degrees of freedom were again below 4 (df = 1.9) and the results were unreliable. The heterogeneity statistics indicated substantial heterogeneity (Q = 9.3, I2 = 78.5%, τ2 = 0.07). The restricted maximum likelihood procedure yielded similar results (ES = −0.06, 95% CI = [−0.57, 0.46], Q = 6.1, I2 = 64.3%, τ2 = 0.03). The main finding of the present review is that there are surprisingly few quantitative studies exploring the effects of changes to adult/child ratio and group size in ECEC on measures of process quality and on child outcomes. The overall quality of the included studies was low, and only two randomised studies were used in the meta‐analysis. The risk of bias in the majority of included studies was high, also in studies used in the meta‐analysis. Due to the limited number of studies that could be used in the data synthesis, we were unable to explore the effects of adult/child ratio and group size separately. No study that examined the effects of changes of the adult/child‐ratio and/or group size on socio‐emotional child outcomes could be included in the meta‐analysis. No high quality study examined the effects of large changes in adult/child ratio and group size on measures of process quality, or explored effects for children younger than 2 years. We included few studies (3) in the meta‐analysis that investigated measures of language and literacy and results for these outcomes were inconclusive. In one specification, we found a small statistically significant effect on process quality, suggesting that fewer children per adult and smaller group sizes do increase the process quality in ECEC. Caution regarding the interpretation must be exerted due to the heterogeneity of the study designs, the limited number of studies, and the generally high risk of bias within the included studies. Results of the present review have implications for both research and practice. First, findings from the present review tentatively support the theoretical hypothesis that lower adult/child ratios (fewer children per adult) and smaller group sizes beneficially influence process quality in ECEC. This hypothesis is reflected in the existence of standards and regulation on the minimum requirements regarding adult/child ratios and maximum group size in ECEC. However, the research literature to date provides little guidance on what the appropriate adult/child ratios and group sizes are. Second, findings from the present review may be seen as a testimony to the urgent need for more contemporary high‐quality research exploring the effects of changes in adult/child ratio and group size in ECEC on measures of process quality and child developmental and socio‐emotional outcomes.