Abstract: This paper investigated how to calculate clients' credit limit applying structural models.Firstly,it derived a structural model including noise which were consistent with china local financial market and the bank's practice of credit risk management.Then,a new method to calculate clients credit limit was brought forward.Finally,a case study was put forward.We found that:(1)When other factors were fixed,the probability of default(PD) was increased monotonously with the increase of default-trigger value;(2) it was useful to increase banks' client credit limit when the noisy information was improved;(3) in comparison with banks actual method,its result was more cautious and more informatory.

Key words: satructural models, noisyinformation, credit risk, default probability