It is well established that cell-free infection with human T-cell leukemia virus type 1 (HTLV-1) is less efficient than that with other retroviruses, though the specific infectivities of only a limited number of HTLV-1 isolates have been quantified. Earlier work indicated that a post-entry step in the infectious cycle accounted for the poor cell-free infectivity of HTLV-1. To determine whether variations in the pol gene sequence correlated with virus infectivity, we sequenced and phenotypically tested pol genes from a variety of HTLV-1 isolates derived from primary sources, transformed cell lines, and molecular clones. The pol genes and deduced amino acid sequences from 23 proviruses were sequenced and compared with 14 previously published sequences, revealing a limited number of amino acid variations among isolates. The variations appeared to be randomly dispersed among primary isolates and proviruses from cell lines and molecular clones. In addition, there was no correlation between reverse transcriptase sequence and the disease phenotype of the original source of the virus isolate. HTLV-1 pol gene fragments encoding reverse transcriptase were amplified from a variety of isolates and were subcloned into HTLV-1 vectors for both single-cycle infection and spreading-infection assays. Vectors carrying pol genes that matched the consensus sequence had the highest titers, and those with the largest number of variations from the consensus had the lowest titers. The molecular clone from CS-1 cells had four amino acid differences from the consensus sequence and yielded infectious titers that were approximately eight times lower than those of vectors encoding a consensus reverse transcriptase.