In this study we investigated the performance of a displaced-beam small aperture scintillometer (DBSAS) - operated over a path length of 112 m - under stable conditions using data gathered during the CASES-99 experiment in Kansas, USA. The DBSAS has the advantage over the eddy covariance method that it can determine fluxes of sensible heat and momentum close to the surface and/or over short (< 1 minute) averaging intervals. Both aspects are of importance in the often shallow and non-stationary stable boundary layer (SBL). From raw DBSAS measurements the dissipation rate of turbulent kinetic energy and the structure parameter of temperature can be deduced, which follow Monin-Obukhov similarity theory (MOST) to give the friction velocity, u* the temperature scale, T* and subsequently the sensible heat flux, H. All these variables were compared with eddy covariance data for 10-minute time averages. Systematic errors were found for the DBSAS u*; i.e. overestimation for low u* values and underestimation for high u* values. It seems these errors cannot be attributed to the use of MOST, since they are already present in the dissipation rate data. In determining the dissipation rate, a form of the temperature spectrum in the dissipation range has to be assumed. The way the DBSAS weights this spectrum will be discussed. Furthermore, the sensitivity of the DBSAS method to small off-sets in instrumental parameters will be discussed. It was found that an adjustment of the beam displacement distance, d that is within the accuracy at which d is determined, removes some of the systematic errors. This adjustment is presented as a working hypothesis, not a general solution.