The association between alcohol consumption and CRP was evaluated in a longitudinal study among 72 middle aged adults for whom 572 repeated measurements of CRP were obtained. Short- and long-term alcohol consumption were analyzed using moving windows of average alcohol consumption levels to establish a dose-response relationship. To control for intra-individual dependence of measurements, generalised estimating equations were used. Average alcohol consumption during the last 12 months showed a clear U-shaped relationship with CRP-levels, with lowest levels found for moderate alcohol consumption of less than 16 g/day. For short-term alcohol consumption no association was found.