The impact of an ongoing loss and fragmentation of reedbeds on reed-inhabiting passerine birds has received much attention in recent years. In addition, global commercial reed harvesting is another important anthropogenic factor threatening the existence of passerine birds. According to previous studies, large-scale winter reed cutting has greatly reduced the number of reed-inhabiting passerine birds across the world. Thus, a new reed harvesting method called 'mosaic reed cutting' has been introduced in many natural reserves to reduce the impact that reed harvesting has on the reed-inhabiting birds. However, the precise impact of the mosaic reed cutting on the habitat use by passerine birds, especially the nest-site selection, has not been well studied. Reed parrotbill (Paradoxornis heudei) is an endemic reedbed-inhabiting bird that is native to East China, and it is currently listed as "near threatened" mainly because of the loss of the wintering and breeding habitats due to the ongoing over-harvesting. In this study, we chose the Yellow River Delta Nature Reserve as a study site to assess the effect of mosaic reed cutting on the nest-site selection and habitat use by reed parrotbills. As habitats can be destroyed by reed harvesting on different spatial scales, we measured 17 variables to determine the nest-site selection of reed parrotbills in the microhabitat (eight variables), territorial patch (nine variables), and full model scales. In total, 141 nests were found and the nest-site characteristics of all the nests were measured during 2008 and 2011-2012. One unused control plot was randomly chosen for each nest and the corresponding habitat variables were also measured. We investigated the nest-site selection of reed parrotbills using binomial logistic regression that was based on two different levels, microhabitat nest-site and nesting patch. The relevant factors for each level were selected based on the Akaike's Information Criterion. The results showed that the nest-site selection of reed parrotbills depended largely on the scales of both microhabitat and nesting patch. However, microhabitat appeared to exert a stronger impact than the nesting patch. In the microhabitat, the nest-site selection positively correlated to the density of dry reed with stems 1.5-2.5 m high, but it was negatively affected by the density of green reed with stems over 2.5 m high, the density of cattail (Typha orientalis), and the visibility of the nest. In nesting patch, the nest-site selection was positively associated with the percentage of dry reed-patch area and vegetation cover, as well as the reed habitat type. However, by combining the two levels, the ultimate factors included all of the above, except for the percentage of dry reed-patch area and reed habitat type. Therefore, reed parrotbills preferred to nest on higher-density dry reed with stems 1.5-2.5 m high, in the microhabitat with lower nest visibility, higher percentage of dry reed-patch area, and habitats with higher vegetation coverage of nesting patch. Thus, 78.7% (n = 141) of birds built their nests in mixed habitats that included both dry and green reed stems and the percentage of dry reeds exceeded 50% in 36.9% of the nests; 53.2% (n = 111) of birds built their nests on substrates formed by both dry and green reeds, and the percentage of dry reeds in these substrates exceeded 50% in 22.3% of the nests. Thus, we concluded that the dry reed left during the mosaic reed harvesting is critical to the nest-site selection and nesting of reed parrotbills, in which both the density of dry reed stems in the microhabitat and the percentage of dry reed-patch area in nesting patch can be significantly reduced by extensive reed harvesting. We further recommended that future reed management should ensure that the reedbed retains at least 20% of dry reed to provide a sustainable nesting habitat for reed parrotbills.