The Internet has been a real revolution that is opening up interesting research possibilities for social scientists. As a survey method, it has obvious advantages that deserve to be explored for election canvassing. On the occasion of the 2010 rector elections at the University of Valencia, an online scientific survey of an electoral process was conducted for the first time in Spain. The survey clearly showed the general trends of the election and the ranking of the candidates, influenced the campaign strategies of the candidates and allowed us to confirm two hypotheses about the behaviour of the voters that we thought a priori were going to occur: the effects of gender and centre of affiliation. The post-stratification and calibration techniques employed, however, were not able to sufficiently correct for the survey's non-response bias: the predictions systematically underestimated the results of the winning candidate.