Internet advertising has grown substantially, and represents the second-largest destination of advertising dollars in the USA. While those data are more difficult to find in other countries, it is expected that the future is similarly bright for the online advertising industry world-wide. This study proposed and tested a causal model that predicts (1) both intentions to return to the host site and recommend the site to others and (2) performance in successfully completing information retrieval tasks on a website. Those two ultimate dependent variables are, in turn, predicted by a user?s feelings of irritation and ad intrusiveness, which are, in turn, predicted by attitude toward the ad and repeated exposure of the ad. In a study involving 420 students, six of the eleven hypothesized relationships were supported: attitude toward the ad predicted intrusiveness; intrusiveness predicted irritation, intentions, and task performance; and both irritation and performance predicted intentions. The model explains 50% of the variance in irritation, 5% of the variance in intrusiveness, 1% of the variance in performance, and 21% of the variance in intentions, providing mixed support for the model.