Background: Beijing's 2024 school physical education reforms have substantially increased activity demands on children, yet population-level genotype data for Chinese schoolchildren remain scarce. Objectives: To characterize ACTN3 R577X, MTHFR C677T, MTHFR A1298C, and MTRR A66G polymorphisms in 564 Beijing schoolchildren; test Hardy-Weinberg equilibrium (HWE); compare allele frequencies with East Asian reference populations; and perform multi-locus folate metabolism risk stratification. Methods: Cross-sectional study of 564 primary and secondary school students (\mbox{265 males}, 299 females) in Beijing. Genotyping was performed by PCR and Sanger sequencing using saliva-derived DNA. HWE was tested via $\chi^2$ goodness-of-fit. Allele frequencies were compared with 1000 Genomes CHB ($n = 103$) and gnomAD EAS ($n \approx 2{,}604$) using chi-squared or Fisher's exact tests. Multi-locus risk scores (0--6 alleles) were computed across three folate-pathway loci. Linkage disequilibrium (LD) and haplotype frequencies were estimated using the EM algorithm. Results: All four loci were in HWE (all $P > 0.05$). Genotype frequencies were: ACTN3 R577X---RR 31.56%, RX 49.11%, XX 19.33%; MTHFR C677T---CC 18.62%, CT 46.99%, TT 34.40%; MTHFR A1298C---AA 71.81%, AC 26.42%, CC 1.77%; MTRR A66G---AA 56.21%, AG 35.82%, GG 7.98%. No significant sex differences were observed at any locus (all $P > 0.05$). Allele frequencies were consistent with East Asian reference populations. Multi-locus risk stratification showed that 70.21% of students carried $\geq$2 risk alleles; 4.79% carried $\geq$4. LD between MTHFR C677T and A1298C was moderate to strong ($|D'| = 0.97$, $r^2 = 0.23$). The double-mutant T-C haplotype was rare (frequency $= 0.002$). Conclusions: This study provides population-level genotype baseline data for exercise- and folate metabolism-related polymorphisms in Beijing schoolchildren. The moderate-to-high prevalence of folate-metabolism risk variants highlights the need for population-level nutritional attention, and the diverse ACTN3 distribution provides a genetic basis for differentiated physical education approaches. These multi-locus frequency data establish a baseline for future genotype-phenotype studies in school-aged populations.