A graduate level research methods and statistics course offered on the World-Wide Web was evaluated relative to the traditional. With their consent, course members were randomly assigned to the two versions of the course for the first block of sessions. For the second block of sessions the groups crossed over to access the alternative version of the course. Quantitative and qualitative outcome data were collected to sample cognitive and affective domains. Improvements in knowledge and reductions in anxiety were apparent following both versions, with no significant differences between versions being detected. Analysis of course member comments indicated less satisfaction with the teaching input on the web-based version but more satisfaction with the peer collaboration that was stimulated. An activity theory framework is applied in conceptualising the findings and generating recommendations for further course development and evaluation.