In infectious HIV-1 particles, the capsid protein (CA) forms a cone-shaped shell called the capsid, which encases the viral ribonucleoprotein complex (vRNP). Following cellular entry, the capsid is disassembled through a poorly understood process referred to as uncoating, which is required to release the reverse transcribed HIV-1 genome for integration into host chromatin. Whereas single virus imaging using indirect CA labeling techniques suggested uncoating to occur in the cytoplasm or at the nuclear pore, a recent study using eGFP-tagged CA reported uncoating in the nucleus. To delineate the HIV-1 uncoating site, we investigated the mechanism of eGFP-tagged CA incorporation into capsids and the utility of this fluorescent marker for visualizing HIV-1 uncoating. We find that virion incorporated eGFP-tagged CA is effectively excluded from the capsid shell, and that a subset of the tagged CA is vRNP associated. These results thus imply that eGFP-tagged CA is not a direct marker for capsid uncoating. We further show that native CA co-immunoprecipitates with vRNP components, providing a basis for retention of eGFP-tagged and untagged CA by sub-viral complexes in the nucleus. Moreover, we find that functional viral replication complexes become accessible to integrase-interacting host factors at the nuclear pore, leading to inhibition of infection and demonstrating capsid permeabilization prior to nuclear import. Finally, we find that HIV-1 cores containing a mixture of wild-type and mutant CA interact differently with cytoplasmic versus nuclear pools of the CA-binding host cofactor CPSF6. Our results suggest that capsid remodeling (including a loss of capsid integrity) is the predominant pathway for HIV-1 nuclear entry and provide new insights into the mechanism of CA retention in the nucleus via interaction with vRNP components.