The distribution and frequency of single nucleotide polymorphisms (SNPs) can help to understand changes associated with characteristics of interest. We aimed to evaluate nucleotide diversity in six genes involved in starch biosynthesis in cassava using a panel of 96 unrelated accessions. The genes were sequenced, aligned, and used to obtain values for nucleotide diversity (π), segregating sites (θ), Tajima's D test, and neighbor-joining (NJ) clustering. On average, one SNP per 147 and 171 bp was identified in exon and intron regions, respectively. Thirteen heterozygous loci were found. Three of seven SNPs in the exon region resulted in non-synonymous replacement or four synonymous substitutions. However, no associations were noted between SNPs and root dry-matter content. The parameter π ranged from 0.0001 (granule bound starch synthase I) to 0.0033 (α-amylase), averaging 0.0011, while θ ranged from 0.00014 (starch branching enzyme) to 0.00584 (starch synthase I), averaging 0.002353. The θ diversity value was typically double that of the π. Results of the D test did not suggest any evidence of deviance of neutrality in these genes. Among the evaluated accession, 82/96 were clustered using the NJ method but without a clear separation of the root dry-matter content, root pulp coloration, and classification of the cyanogenic compound content. High variation in genes of the starch biosynthetic pathway can be used to identify associations with the functional properties of starch for the use of polymorphisms for selection purposes.