Conclusive evidence-based research has shown that circumcision reduces the risk of HIV transmission via heterosexual intercourse, whilst ongoing studies are investigating similar effects via homosexual transmissions and the results are equivocal. Few acceptability studies regarding circumcision were conducted among men who have sex with men (MSM). In this cross-sectional study, a total of 307 MSM were recruited by snowball sampling and were interviewed anonymously by some peer field workers in Yangzhou, China. Amongst all uncircumcised participants (93.4% of all participants were uncircumcised), the willingness to be circumcised increased from 8.1% to 30.7%, before and after the participants were briefed about a hypothetical potential benefit of a 50% risk reduction of circumcision in preventing HIV transmission among MSM. In the multivariate analysis, perception of overly long foreskin (odds ratio [OR] = 6.04), unprotected sexual intercourse with male regular sex partners in the last six months (OR = 2.04), and seeing no chance for contracting HIV in the next 12 months (OR = 0.54) were significantly associated with conditional willingness for circumcision. Adjusting for these variables, other significant factors were identified by multiple logistic regression analysis, including number of perceived disadvantages for having overly long foreskin (adjusted OR = 2.60), variables that were derived from the Theory of Planned Behaviors (TPB), and having some circumcised MSM peers (adjusted OR = 0.45-4.38). Some risk compensation behaviors however, may be practiced by 15.9% of the MSM who were willing to undergo circumcision. The acceptability would increase slightly with the effect size of circumcision in protecting MSM from HIV transmission via homosexual intercourse. However, it was only around 30%, even if circumcision could result in a large (50%) risk reduction in HIV transmission among MSM. If future studies can establish efficacy of circumcision, relevant promotion programs need to guard against risk compensation, though the magnitude of risk compensation may be moderate.