OBJECTIVE: Using clinical isolates from a recent passive immunization trial with antibody 2G12, we probed the capacity of frequently used neutralization formats - the primary peripheral blood mononuclear cell (PBMC)-based and the TZM-bl cell-based assay systems - to predict in-vivo activity of 2G12. DESIGN: Antibodies and entry inhibitors of established efficacy were used to neutralize HIV-1 isolates in different in-vitro assay setups. METHODS: Single round infection with Env-pseudotyped and multiple round infection with replication-competent virus was studied on PBMCs and a variety of engineered cell lines. RESULTS: Six out of 12 isolates with high sensitivity to 2G12 in the replication-competent PBMC assay lacked sensitivity to the monoclonal antibody in the env-pseudotype TZM-bl assay. Outcome of passive immunization with 2G12 corroborated the PBMC-assay in-vitro data, as escape mutations to 2G12 emerged, proving the monoclonal antibody's impact on HIV in vivo. Failure to inhibit pseudotype infection of TZM-bl was not due to sequence differences or the pseudotype infection per se, as infection of PBMCs with the identical pseudotyped viruses was sensitive to 2G12 inhibition. Similar shifts in efficacy, though less extreme, were noted for other neutralizing antibodies and inhibitors. Exploration of causes for the observed differences between assay systems revealed that both target cell and virus producer properties influence sensitivity of virus entry to inhibition. CONCLUSION: Our observation that neutralization assay systems employing engineered reporter cell lines can miss in-vivo relevant neutralizing activities strongly argues that preclinical assessment should not be restricted to a single assay type.