2-Keto-l-gulonic acid (the precursor of vitamin C) is bio-converted from l-sorbose by mixed fermentation of Ketogulonicigenium vulgare and a helper strain. The helper strain promotes the conversion of 2-KLG by enhancing the growth of K. vulgare, but its growth is greatly inhibited by high concentration of l-sorbose, which consequently influence the 2-KLG production. The aim of this study is to obtain l-sorbose-tolerant helper strain (LHS) by experimental evolution for reduced l-sorbose-inhibition-effect and enhanced 2-KLG productivity in high concentration of l-sorbose. After three steps screening by using our devised screening strategy, three strains (i.e., Bc 21, Bc 47, Bc 50) with high resistance to high concentration of l-sorbose were obtained. The fermentation tests by co-culturing Bc 21 and K. vulgare 418 showed that the production of 2-KLG was increased by 17.9% in 11% l-sorbose medium than that in 8% after 55 h of fermentation and the conversion rate was 89.5%. The results suggested that Bc 21 could be an ideal helper strain for 2-KLG production under high concentration of l-sorbose and demonstrated the feasibility of using experimental evolution to breed LHS for vitamin C production.