The sources of spontaneous point mutations in genomes are diverse, including DNA damages and errors introduced by the polymerase during replication. The resulting mutation rate also depends on the counteracting action of DNA repair mechanisms, with mutator phenotypes appearing constantly and playing a role during phases of rapid evolution in nature and in the laboratory. Here, we use the gram-positive model bacterium Bacillus subtilis to jointly assess the respective contributions to the mutation rate of DNA polymerase nucleotide selectivity, proofreading, and mismatch repair (MMR). For this purpose, we constructed and analysed several conditional hypermutators with a proofreading-deficient allele of polC and/or a deficient allele of mutL. By covering a wide range of mutation rates and displaying contrasted mutation profiles, these conditional hypermutators enrich the B. subtilis synthetic biology toolbox for directed evolution. Analysis of their mutation profiles in light of several mathematical models exposes the difficulties of interpreting apparent probabilities of error correction that stem from aggregating possibly heterogeneous subclasses of mutations into counts, and from unknowns on the components of the mutation profiles in the presence of the repair systems. Aware of these difficulties, the analysis strongly suggests that proofreading is needed to avoid partial saturation of the MMR in B. subtilis and that an inherent effect of proofreading is to skew the net polymerase error rates by reinforcing intrinsic biases of nucleotide selectivity.