Peppers need to be characterized for the in situ conservation of species that might be important for breeding purposes. In this study, we characterized the diversity of 23 pepper genotypes through eight qualitative and 12 quantitative descriptors. The pepper samples were collected from the city of Manaus and in the municipalities of Rio Preto da Eva and Iranduba in the state of Amazonas. We performed descriptive analysis, analysis of variance, Pearson’s correlation analysis, and principal component analysis (PCA) and used the unweighted pair group method with arithmetic mean (UPGMA) to analyze the data. The differences between the genotypes for all descriptors evaluated were significant. Among the genotypes collected, four pepper species were identified, including Capsicum chinense (14 genotypes), Capsicum frutescens (two genotypes), Capsicum baccatum (one genotype), and Capsicum annuum (four genotypes), and two other genotypes were obtained with unidentified species. The PCA and UPGMA clustering methods could be used to efficiently differentiate between the genotypes by applying quantitative and qualitative descriptors of fruits and seeds. These methods helped to identify different genotypes (G1 and G15) and separate them into distinct groups associated with the distribution of species based on quantitative data only. Overall, our findings suggested that the genetic variability in Capsicum available in the State of Amazonas might be useful in breeding programs.