We have observed previously that DNA template-directed polymerization by the type 1 human immunodeficiency virus reverse transcriptase is error-prone for single-nucleotide substitution, addition and deletion errors at homopolymeric sequences. We have also noted strong termination of processive synthesis at these positions (Bebenek, K., Abbotts, J., Roberts, J. D., Wilson, S. H., and Kunkel, T. A. (1989) J. Biol. Chem. 264, 16948-16956). Here we have tested three models to explain errors at these hot spots: template-primer misalignment for deletion errors, and dislocation and direct miscoding for substitution errors. The approach involves introducing single-nucleotide changes within or flanking the homopolymeric hot spots and examining the effects that these changes have on human immunodeficiency virus type 1 (HIV-1) reverse transcriptase error rate, error specificity, and termination probability. The results obtained suggest that single-nucleotide deletion errors in homopolymeric runs result from template-primer misalignment and that both direct miscoding and template-primer dislocation contribute to the base substitution hot spots. The data also suggest that base substitution errors at one position can be templated by the preceding nucleotide or either of the next two nucleotides. Frameshift error rates at homopolymeric sites were affected by changes in the sequences flanking the runs, including single-nucleotide differences in the single-stranded template strand and in the double-stranded primer region as many as six nucleotides distant from the hot spot. Both increases and decreases in frameshift fidelity were observed, and most of these correlated with concomitant increases or decreases in the probability that HIV-1 reverse transcriptase terminated processive synthesis within the run. These data provide further support for a relationship between the frameshift fidelity and the processivity of DNA-dependent DNA synthesis by HIV-1 reverse transcriptase.