Nanometer-sized metal and semiconductor particles possess novel properties. To fully realize their potential, these nanoparticles need to be fabricated into ordered arrays or predesigned structures. A promising nanoparticle fabrication method is coupled surface passivation and self-assembly of surfactant-coated nanoparticles. Due to the empirical procedure and partially satisfactory results, this method still represents a major challenge to date and its refinement can benefit from fundamental understanding. Existing evidences suggest that the self-assembly of surfactant-coated nanoparticles is induced by surfactant-modified interparticle interactions and follows an intrinsic road map such that short one-dimensional (1D) chain arrays of nanoparticles occur first as a stable intermediate before further assembly takes place to form higher dimensional close-packed superlattices. Here we report a study employing fundamental analyses and Brownian dynamics simulations to elucidate the underlying pair interaction potential that drives the nanoparticle self-assembly via 1D arrays. We find that a pair potential which has a longer-ranged repulsion and reflects the effects of surfactant chain interdigitation on the dynamics is effective in producing and stabilizing nanoparticle chain arrays. The resultant potential energy surface is isotropic for dispersed nanoparticles but becomes anisotropic to favor the growth of linear chain arrays when self-assembly starts.