We found that hNAP1 bound to H2A-H2B facilitates proper nucleosome formation by inhibiting random binding of H2A-H2B to DNA and that the deletion of the C-terminal acidic domain (CTAD) of human NAP1, which is intrinsically disordered, caused complete loss of inhibition of its random binding to DNA. In addition the hNAP1 CTAD alone specifically bound to H2A-H2B with high affinity, which is close to the whole hNAP1 dimer. However, the mode of interaction between NAP1 and H2A-H2B remains elusive. Initially we determined the solution structure of isolated H2A-H2B heterodimer by NMR together with MD calculations and showed that both H2A and H2B contain a histone fold, which is essentially the same as the corresponding parts in the nucleosome. However the N- and C-terminal helices outside the histone fold of H2A, two helices found in the nucleosome are entirely disordered in the isolated H2A-H2B. Subsequently, we examined the interaction with human NAP1 (hNAP1). The disordered CTAD of hNAP1 specifically bound to H2A-H2B like a flexible extended string; an aspartic acid rich CTAD segment binds to a H2A-H2B basic patch inducing the formation of the H2A N-terminal α helix as found in the nucleosome, where αN is stabilize by DNA, suggesting that the aspartic acid rich segement behaves like DNA. In addition a tyrosine containing CTAD segment recognizes a small hydrophobic patch on H2A-H2B. These binding modes were established by mutagenesis experiments involving the CTAD and fragments themselves. In addition DNA nonspecifically binds to the H2A-H2B core surface, suggesting DNA bending in the minor groove is imposed by H2A-H2B; DNA in the complex is easily replaced by CTAD-binding, allowing hNAP1 to inhibit random transfer of H2A-H2B to DNA.