Characterization of multiple sites in a single gene that are important in biological phenotypes is challenging due to the difficulty to generate many mutants representing all or a majority of combinations of mutations in the gene. Using the HIV-1 env and pol genes as templates, four random libraries were generated representing different combinations of mutations introduced by up to 36 mutagenesis primers in a single assay. Over 86% of the clones contained mutations and the mutants tended to have single or fewer mutations in the libraries. When protein size was used as a screening marker, all identified clones contained at least 2 mutations and up to 12 mutations were detected in a single clone. Nearly all mutant clones in each library contained unique mutations, indicating that mutants in the library were generated at random. Closely related mutations which were overlapped by neighboring mutagenesis primers were often introduced in this system. Analysis of the env library showed that some potential N-linked glycosylation sites did not increase the Env molecular mass significantly, suggesting they were not used for glycosylation or only limited carbohydrate moieties were added at these sites. This novel method can serve as a powerful tool to study the biological phenotypes of genes whose functions are determined by multiple sites.