Many breast cancer patients currently turn to Internet-based education and support to help them cope with their illness. This study explores the role of training in influencing how patients use a particular Interactive Cancer Communication System (ICCS) over time and also examines what pre-test characteristics predict which people are most likely to opt in or out of training in the first place. With use of pre-test survey and unobtrusive individual records of ICCS system use data (N = 216), nonparametric tests revealed that only having a later stage of cancer predicted whether or not patients participated in training. Results indicated that participating in training was a significant predictor of higher levels of using the CHESS system. In particular, the repeated measures analysis of covariance found the significant interaction as well as main effect of group (i.e., training vs. no training) and time (i.e., individual's CHESS usages at different times) in interactive and information CHESS services, suggesting that 1) the training group has a higher level of usage than the no training group, 2) both of the groups' usage decreased over time, and 3) these joint patterns hold over time. Practical guidelines for future ICCS campaign implementation are discussed.