Adaptive laboratory evolution (ALE) is used to find causal mutations that underlie improved strain performance under the applied selection pressure. ALE studies have revealed that mutator populations tend to outcompete their non-mutator counterparts following the evolutionary trajectory. Among them, inactivated mutator cells, characterize d by a dysfunctional methyl-mismatch repair system, are frequently found in ALE experiments. Here, we examined inactivation as an approach to facilitate ALE of The wild-type MG1655 and knock-out derivative (Δ) were evolved in parallel for 800 generations on lactate or glycerol minimal media in a serial-transfer experiment. Whole-genome re-sequencing of each lineage at 100-generation intervals revealed that (1) mutations emerge rapidly in the Δ compared to in the wild-type strain; (2) mutations were more than fourfold higher in the Δ strain at the end-point populations compared to the wild-type strain; and (3) a significant number of random mutations accumulated in the Δ strains. We then measured the fitness of the end-point populations on an array of non-adaptive carbon sources. Interestingly, collateral fitness increases on non-adaptive carbon sources were more pronounced in the Δ strains than the parental strain. Fitness measurement of single mutants revealed that the collateral fitness increase seen in the mutator lineages can be attributed to a pool of random mutations. Together, this study demonstrates that short-term mutator ALE extensively expands possible genotype space, resulting in versatile bacteria with elevated fitness levels across various carbon sources.