The conventional system of paper checkbooks with multiple checks carries a significant risk of default because no form of authorization is required when individual checks are written, although implicitly each checkbook is authorized when it is issued to the payer. Nonetheless, checks are a less expensive payment method than credit cards, so they are suitable for a high-credit business environment. To make up for the defect in the paper check sysem, whereby checks likely to default can be issued without authorization, the authors have devised an agent-based electronic check system known as SafeCheck that can monitor the situation and block the issuance of nonallowable checks in a distributed manner. Three types of service are allowed, depending upon the check issuer's credibility. Members of the top credit class require authorization only for each checkbook. Members of the second-level credit class require authorization for each check. Members of the third-level credit class are allowed to issue checks only within their checking account balance. The bank can dynamically adjust the credit level depending upon the record of defaults. The SafeCheck System consists of three agents: Checkbook Agent at the check issuer's site, Check-Receipt Agent at the check receiver's site, and the bank's Control Agents at the check issuer's and receiver's banks respectively. For security purposes, SafeCheck has public key cryptography, digital signatures, and certificate schemes like those of the SET protocol for credit cards. The essence of a checkbook agent can be stored in the IC card.