Hepatitis C virus (HCV) is classified into seven phylogenetically distinct genotypes, which are further subdivided into related subtypes. Accurate assignment of genotype/subtype is mandatory in the era of directly acting antivirals. Several molecular methods are available for HCV genotyping; however, a relevant number of samples with indeterminate, mixed, or unspecified subtype results, or even with misclassified genotypes, may occur. Using NS5B direct (DS) and ultra-deep pyrosequencing (UDPS), we have tested 43 samples, which resulted in genotype 1 unsubtyped (n = 17), mixed infection (n = 17), or indeterminate (n = 9) with the Abbott RealTime HCV Genotype II assay. Genotype 1 was confirmed in 14/17 samples (82%): eight resulted in subtype 1b, and five resulted in subtype 1a with both DS and UDPS, while one was classified as subtype 1e by DS and mixed infection (1e + 1a) by UDPS. Three of seventeen genotype 1 samples resulted in genotype 3h with both sequencing approaches. Only one mixed infection was confirmed by UDPS (4d + 1a), while in 88% of cases a single component of the mixture was detected (five genotype 1a, four genotype 1b, two genotype 3a, two genotype 4m, and two genotype 4d); 44% of indeterminate samples resulted genotype 2c by both DS and UDPS, 22% resulted genotype 3a; one indeterminate sample by Abbott resulted in genotype 4d, one resulted in genotype 6n, and one was classified as subtype 3a by DS, and resulted mixed infection (3a + 3h) by UDPS. The concordance between DS and UDPS was 94%, 88%, and 89% for genotype 1, co-infection, and indeterminate results, respectively. UDPS should be considered very useful to resolve ambiguous HCV genotyping results.