We designed a single nucleotide primer extension (SNaPshot) assay for Pneumocystis jirovecii genotyping, targeting mt85 SNP of the mitochondrial large subunit ribosomal RNA locus, to improve minority allele detection. We then analyzed 133 consecutive bronchoalveolar lavage (BAL) fluids tested positive for P. jirovecii DNA by quantitative real-time PCR, obtained from two hospitals in different locations (Hospital 1 [n = 95] and Hospital 2 [n = 38]). We detected three different alleles, either singly (mt85C: 39.1%; mt85T: 24.1%; mt85A: 9.8%) or together (27%), and an association between P. jirovecii mt85 genotype and the patient's place of hospitalization (p = 0.011). The lowest fungal loads (median = 0.82 × 10(3) copies/μl; range: 15-11 × 10(3) ) were associated with mt85A and the highest (median = 1.4 × 10(6) copies/μl; range: 17 × 10(3) -1.3 × 10(7) ) with mt85CTA (p = 0.010). The ratios of the various alleles differed between the 36 mixed-genotype samples. In tests of serial BALs (median: 20 d; range 4-525) from six patients with mixed genotypes, allele ratio changes were observed five times and genotype replacement once. Therefore, allele ratio changes seem more frequent than genotype replacement when using a SNaPshot assay more sensitive for detecting minority alleles than Sanger sequencing. Moreover, because microscopy detects only high fungal loads, the selection of microscopy-positive samples may miss genotypes associated with low loads.