Interaction is one of the controversial topics in e-learning literature. Overall body of knowledge that has accumulated over the past decade seemed inconclusive and needs critical analyses. To find an answer to the perplexing issue, this study presents a path analysis model to investigate the effects of interaction and dialog on e-learning outcomes and satisfaction. The path analysis model we present here differs from all existing studies in that the path model introduced a new variable, dialog. The concept of dialog is suggested by Moore, but it was not empirically test until now. Based on the review of related research, five hypotheses were developed. The structural equations also show that about forty two percent of e-learning satisfaction levels can be explained by three endogenous variables and one exogenous variable (learning outcomes). The level of student satisfaction in e-learning hinges on the facilitating role of the instructor, the interaction among students, and the dialogue between the instructor and students, in the order of the magnitude of path coefficients.