Digital homogeneous non-enzymatic immunosorbent assay (digital Ho-Non ELISA) is a new class of digital immunoassay that enables highly sensitive quantification of biomolecules using a simple protocol. In digital Ho-Non ELISA, nanoparticles are tethered onto the surface of femtoliter reactors via captured target molecules. The tethered particles capturing target molecules are identified as those showing a confined Brownian motion with root-mean-square displacement (RMSD) values in a defined range. The present work aims to improve the specificity to discriminate tethered particles via single-target molecules from non-specifically immobilized particles by analyzing two nanoparticle parameters. First, in order to suppress the broadening of RMSD due to the heterogeneity of bead size, we corrected the RMSD with the fluorescence intensity of the beads. Second, focusing on the shape of Brownian motion in the x–y trajectory, we classified motion patterns by aspect ratio of the trajectory. By using multiparameter single-particle motion analysis with corrected RMSD and aspect ratio, a 3.9-fold enhanced sensitivity in PSA assay was achieved compared to the conventional RMSD analysis approach. This new strategy would increase the potential of digital immunoassays.