The effect of dust particle size on the distribution and bioaccessibility of flame retardants (FRs) in indoor dust remains unclear. In this study, we analyzed 20 FRs (including 6 organophosphate flame retardants (OPFRs), 8 polybrominated diphenyl ethers (PBDEs), 4 novel brominated flame retardants (NBFRs), and 2 dechlorane plus (DPs)) in composite dust samples from offices, public microenvironments (PME), and cars in Nanjing, China. Each composite sample (one per microenvironment) was separated into 6 size fractions (F1–F6: 200–2000 μm, 150–200 μm, 100–150 μm, 63–100 μm, 43–63 μm, and <43 μm). FRs concentrations were the highest in car dust, being 16 and 6 times higher than those in offices and PME. The distribution of FRs in different size fractions was Kow-dependent and affected by surface area (Log Kow=1–4), total organic carbon (Log Kow=4–9), and FR migration pathways into dust (Log Kow>9). Bioaccessibility of FRs was measured by the physiologically-based extraction test, with OPFR bioaccessibility being 1.8–82% while bioaccessible PBDEs, NBFRs, and DPs were under detection limits due to their high hydrophobicity. The OPFR bioaccessibility in 200–2000 μm fraction was significantly higher than that of <43 μm fraction, but with no difference among the other four fractions. Risk assessment was performed for the most abundant OPFR-tris(2-chloroethyl) phosphate. The average daily dose (ADD) values were the highest for the <43 μm fraction for all three types of dust using total concentrations, but no consistent trend was found among the three types of dust if based on bioaccessible concentrations. Our results indicated that dust size impacted human exposure estimation of FRs due to their variability in distribution and bioaccessibility among different fractions. For future risk assessment, size selection for dust sampling should be standardized and bioaccessibility of FRs should not be overlooked.