Business microdata should be able to measure the employment effect of the shorter working week based on a comparison of firms that have switched to the 35-hour week with those that have remained on 39 hours. Such a comparison should cover firms that are as similar as possible. However, certain questions remain: do we have enough information on these firms to make them comparable or are there also unmeasured microeconomic characteristics that differentiate the two groups? Do the firms have the same capacity to adapt to the shorter working week? Lastly, can we consider that the effects of the shorter working week concern only the companies that have made the transition or have there also been indirect effects on the firms that have kept the 39-hour working week? These are complex questions. We address them first of all by looking at the effects of the shorter working week on production and productivity. For example, given comparable characteristics, the firms that switched to 35 hours under the Aubry I act posted a slight decrease of 3.7% in their total factor productivity, i.e. their capacity to produce for constant quantities of labour and capital, from 1997 to 2000 compared with those still on 39 hours at the end of 2000, whereas the transition to 35 hours should have reduced the working week in these firms by four hours or 10.2%. At the same time, employment in these firms is found to have risen 9.9% compared with the firms that remained on a 39-hour week.