This article gives a detailed analysis of devoicing of the voiced fricatives /v/, /z/, /y/ in two varieties of Standard Dutch: Southern Standard Dutch (as spoken in Belgium) and Northern Standard Dutch (as spoken in The Netherlands). The study is based on archived recordings of radio broadcasts from 1935 to 1993. First, our study shows a divergence between Southern and Northern Standard Dutch in the pronunciation of voiced fricatives in this period. In The Netherlands there is a strong tendency towards devoicing, but in Belgium this tendency is very weak. Second, this study offers insight into the linguistic path of this change: partially voiced compromise variants play an important role, and devoicing is favored in word-initial position. Finally, our study shows the benefits of a retrospective trend study on the basis of radio recordings. In comparison with traditional real-time studies, it offers more insight into the social and linguistic embedding of changes in progress. Its results are also more reliable than those of apparent-time research.