Adequate lying times are essential to the well being of the dairy cow. Hence the assessment of lying behaviour is an important welfare indicator of cows. In robotic milking farms (RMF), restricted access to specific barn areas and night time milking may result in disturbed lying behaviour. The aim of this study was (1) to validate a new Data Logger (DL: 20 x 15 x 61 mm, 23 g, MSR Electronics GmbH, Switzerland) for recording lying bouts via acceleration, and (2) to investigate the lying behaviour of two different groups of cows on four commercial RMF in Switzerland. (1) DLs were programmed with a logging interval of 30 s and attached to the left hind legs of 19 cows. Video recordings from a two-day period were analysed. A total concordance of 99.92 % between video observations and DL readings was obtained. (2) On each farm, 20 cows were previously selected on the basis of herd management data and assigned to two groups of ten animals each. In contrast to individuals of group A, group B cows were characterised by long intervals between milkings, a low milking frequency and a high percentage of night time milking. Both groups had an equal distribution of parity and stage of lactation. Data recorded over a period of 11-14 days were analysed using multifactor ANOVA and regression analysis. Group B cows showed a longer daily lying time than individuals of group A (677 min vs. 638 min, F = 21.64, p = 0.000) and the interval between milkings correlated highly significant with total lying time (r = 0.18 - 0.48). Due to its accuracy, the DL represents an excellent alternative to direct or video observation of the lying behaviour of cows and provides a non-invasive method for assessing the welfare of dairy cows on RMF.