This study examined the mutual influence processes between self-appraisal on the Internet and actual self-appraisal. Specifically, we researched the discrepancies in each self-evaluation, and how the degree of those discrepancies related to mental health. In a forum on the Internet, a survey with a two-wave panel design was conducted on the Web for mothers with pre-school children who were exchanging childcare information. At the time of each investigation, the actual self-appraisal (SA), the reflected self-appraisal (RSA: participants infer how a significant other evaluates them), the reflected self-appraisal on the Net (RSA-N: participants infer how a significant other in the forum in which they participate evaluates them), and mild depression were measured as an index of mental health. The result showed that the discrepancy betweerl SA and RSA-N was significantly larger than the discrepancy between SA and RSA. We further found that the level of the RSA-N score was significantly lower than that of RSA or SA. However, depression was not influenced by the lowness of the RSA-N score or the discrepancy between SA and RSA-N, but was instead influenced by the lowness of the SA score or the discrepancy between SA and RSA. Moreover, Path analysis found the self-process on the Internet. Specifically, Time 1 SA affected Time 2 RSA; in turn, RSA correlated with RSA-N at Time 2. These results suggested that the self-appraisal on the Internet was formed based on actual self-appraisal by using the Internet about an actual problem.