Internet is a very attractive technology for experiments implementation, both in order to reach more diverse and larger samples and as a field of economic research in its own right. This paper reports on an experiment performed both online and in the laboratory, designed so as to strengthen the internal validity of decisions elicited over the Internet. We use the same subject pool, the same monetary stakes and the same decision interface, and randomly assign two group of subjects between the Internet and a traditional University laboratory to compare behavior in a set of social preferences games. This comparison concludes in favor of the reliability of behaviors elicited through the Internet. Our behavioral results contradict the predictions of social distance theory, as we find that subjects allocated to the Internet treatment behave as if they were more altruistic, more trusting, more trustworthy and less risk averse than laboratory subjects. Those findings have practical importance for the growing community of researchers interested in using the Internet as a vehicle for social experiments and bear interesting methodological lessons for social scientists interested in using experiments to research the Internet as a field.