Background/rationale: There is clear evidence that the COVID-19 pandemic had significant negative effects on the learning and development of school-age children in the United States, with disproportionate impacts on children from racially, ethnically, and socioeconomically marginalized groups (Irwin et al., 2022). There is less consistent evidence on the extent to which the pandemic affected younger, preschool-age children (Sparks, 2022; Deoni et al., 2022). There is a clear need for better information on the academic and cognitive development of the country's youngest learners--particularly those from marginalized groups--in the wake of the pandemic. Acelero Learning operates dozens of Head Start centers serving children in four delegate sites across the country: Philadelphia, NJ, Milwaukee, and Las Vegas. Acelero identified a need to better understand whether children in their programs were exhibiting resilience during the pandemic recovery. That is, were their assessment scores across a variety of domains similar to or better than those of demographically similar populations of children attending Head Start before the crisis? (Weiland & Morris, 2022). Research questions: This study answers two questions: 1. To what extent did 3- and 4-year-old children enrolled in Acelero Learning Head Start programs exhibit resilience two years after the start of the pandemic? 2. Did children's gains in academic and cognitive skills during this time vary for different demographic groups, including race and ethnicity, age, gender, and linguistic background? Study sample: Researchers randomly selected 37 centers, 79 classrooms, and 475 children--split across the four sites--to participate in the study in January 2022. The large majority of children in the study were from families with lower incomes or who were experiencing poverty (see Figure 1 for demographics). Measures. In winter and spring 2022, the field-based research team collected direct assessments of children's skills in language, literacy (both print knowledge and phonological awareness), math, and executive functioning (see Box 1 for a complete list of the assessments used). The analytic sample -- 343 -- represents students with assessments in winter and spring. Assessments have been used in other studies of Head Start programs in general and Acelero Learning programs in particular, allowing for pre-pandemic comparisons. We had access to demographics from the Acelero's administrative records. Analytic approach. We used descriptive statistics to examine raw and standardized scores on each assessment. Standardized scores allowed us to directly compare children in this study to earlier samples and to examine whether children made meaningful gains in key learning domains during this period relative to pre-pandemic norms. Box 2 describes the comparison study samples. In contrast, examination of raw scores allowed researchers to describe children's gains in terms of standard deviation units and months of learning (Lipsey et al., 2013). We used multi-level residualized gains models to examine how growth varied for subgroups. Results. Learning gains were fastest for language and phonological awareness (see Table 1). Gains were aligned with what would be expected in math (2.86 months) and somewhat slower than expected in print knowledge (2.01 months). Overall scores and learning gains in language, literacy and executive functioning were comparable to pre-pandemic norms, while math scores were lower (see Figure 2). Math skills scores for the current sample were lower than both the pre-pandemic national Head Start sample and the pre-pandemic Acelero Learning sample; gains in math were similar to the pre-pandemic national Head Start sample but slower than the comparable sample of pre-pandemic Acelero Learning students (see Figure 2). Older children (4 years old in fall) made faster gains in language, phonological awareness, and math skills than younger children, and boys made larger gains in language skills than girls. Children from families with single parents also made larger gains in language than children living in two-parent households. Differences in gains across domains were substantial and ranged from about 1.5 months to 3 months of learning. Implications. Children demonstrated growth in all domains. In language, literacy, and executive functioning, their performance scores and gains were in line with (or faster than) those of similar children in pre-pandemic Acelero Learning programs (Aikens et al., 2017). However, there are areas for further work. Overall scores were lower and growth was slower in math post-pandemic compared with pre-pandemic levels, similar to performance data for school-age children. Further investigation into how teachers are working to support math is needed. There are also inklings from this study that the children perhaps most negatively affected by the pandemic -- boys, slightly older children, and children living with one parent -- are bouncing back.