Conventional assays based on infection of T-cell lymphoblastoid lines with tissue culture-adapted strains of human immunodeficiency virus (HIV) are well established and have been used successfully to discover potent inhibitors of HIV replication. In this report we show that such assays are not easily applied to testing the susceptibilities of clinical HIV isolates to inhibitors because of differences in replication rates and cytotoxicity, thus demonstrating that conventional HIV assays should be used with caution when the zidovudine susceptibility of clinical isolates is assessed. An assay based on plaque reduction in CD4+ HeLa cell monolayers was validated by determining susceptibilities of HIV to a large number of inhibitors in this system. In general, 50% inhibitory doses for HIV type 1 and 2 strains derived from plaque reduction data were in good agreement with susceptibility data obtained by using conventional assays with T-cell lines. The susceptibilities of previously identified zidovudine-resistant HIV isolates to a large group of inhibitors, including nonucleosides, such as interferons and soluble CD4, were tested by using a plaque reduction assay in CD4+ HeLa cells. Surprisingly, an extremely narrow range of cross resistance was observed; cross resistance was limited to nucleoside analogs containing a 3'-azido group. These data point the way to the use of combinations of inhibitors to delay the appearance of drug resistance.