What influences online service quality and how is service quality linked to customer satisfaction? This dissertation addresses these important questions using four alternative operationalizations of the service dimensions: expectations minus perceptions (E-P), a direct measure of the expectations - perceptions gap (direct), the perceptions measure only (P), and expectations minus perceptions adjusted for the absolute level of perceptions P-(E-P). Separate structural equation models are estimated for each measurement operationalization. The study found that E-P variables were not highly reliable and the discriminant validity for all models between service quality predictors was not high. A simple overall measure of service quality outperformed independent service dimensions for all models. The only significant individual predictors were sense of internet community and site customization. This study found that Internet buyers are not greatly sensitive to individual service dimensions which seem to act as indicators of a holistic latent service evaluation variable. Buyers are more sensitive to the transaction outcome than its process. A halo effect across service dimensions was also found. Satisfaction was linked to only cognitive, affective, and conative loyalty but not behavioral loyalty. When word of mouth was used as a proxy loyalty measure satisfaction explained more of the variance than the loyalty variables in all models. In assessing overall service, performance managers can use the overall variable. But given that a halo effect was found in this study, identifying and managing specific service dimensions is difficult.