Wide-field photometry is prone to various degradations, such as atmospheric ex- tinction, varying point spread functions, and aliasing in addition to classical noise sources such as photon, sky background, readout, and thermal noise. While space- borne observations do not suer from atmospheric eects, varying star images over a large sensor and aliasing may seriously impede good results. A measure of the achievable precision of ground-based dierential photometry with the prototype photometer for the BRITE satellite mission is reported, using real sky observa- tions. The data were obtained with the photometer attached to a paramount tracking platform, using the Image Reduction and Analysis Facility Software (IRAF) image reduction and analysis methods as well as the author's own Matlab Code. Special emphasis is placed on the analysis of varying apertures for vary- ing point spread functions, which shows that the accuracy can be improved by taking into account the statistics for each star instead of using a xed aperture. In addition a function is dened, which describes the expected error in terms of instrumental magnitudes, taking into account Poisson distributed noise and mag- nitude independent noise, mainly aliasing. This function is then t to observed data in a two-dimensional least squares sense, providing a calculated aliasing error of 7 millimagnitudes. This function is furthermore rewritten in terms of the stan- dard magnitude B. A maximum magnitude can then be determined for a certain precision, which shows that the Bright Target Explorer (BRITE) can reach a pho- tometric error of 1 millimagnitude for stars with magnitude B &lt; 3:5, assuming the worst case duty cycle of 15 minutes. / <p>Validerat; 20110211 (anonymous)</p>