The French case offers a very valuable opportunity for testing the impact of transportation costs on the individual decision of turnout, because in France, voters can be registered in a municipality other than their residential municipality. Voters with a non-residential registration have to travel great distances in order to cast their ballots compared to voters with a residential registration. Our empirical analysis, based on a unique dataset extracted from the French Census database, uses a selection model to estimate the probability of voting of an individual voter with a non-residential electoral registration and assesses the impact of the distance to the voting municipality on this probability. The analysis shows that distance and in fine the cost of voting have a highly significant impact on electoral turnout: at the median distance (22 km), a 1% increase in distance induces a reduction of 0.01% in turnout in the first round of the 2012 French presidential election and 0.007% in the second round. Moreover, the impact of distance is non-linear: an increase in distance for voters with a short distance to travel is more detrimental to turnout than the same increase for voters who travel great distances. The results are robust to several checks, ranging from analyzing other elections to changes in the estimation method. This analysis provides new insights in the issue of voting cost and its impact on electoral turnout.