Studying RS CVn binaries is challenging, because in addition to spot activity, other effects such as mass transfer between the components and gravitational distortion of their spherical forms may distort their light curves. Such effects can, however, be removed from the data by subtracting a mean light curve phased with the orbital period. We study a quarter of a century of standard Johnson differential V photometry of the RS CVn binary BM CVn. Our main aims are to determine the activity cycles, the rate of surface differential rotation and the rotation period of the active longitudes of BM CVn. The Continuous Period Search (CPS) is applied to the photometry. The changes of the mean and amplitude of the light curves are used to search for activity cycles. The rotation period changes give an estimate of the rate of surface differential rotation. The Kuiper method is applied to the epochs of the primary and secondary minima to search for active longitudes. The photometry reveals the presence of a stable mean light curve (MLC) connected to the orbital period P_orb = 20d.6252 of this binary. We remove this MLC from the original V magnitudes which gives us the corrected V' magnitudes. These two samples of V and V' data are analysed separately with CPS. The fraction of unreliable CPS models decreases when the MLC is removed. The same significant activity cycle of approximately 12.5 years is detected in both V and V' samples. The estimate for the surface differential rotation coefficient, k >= 0.10, is the same for both samples, but the number of unrealistic period estimates decreases after removing the MLC. The same active longitude period of P_al = 20d.511 +- 0d.005 is detected in the V and V' magnitudes. This long-term regularity in the epochs of primary and secondary minima of the light curves is not caused by the MLC. On the contrary, the MLC hampers the detection of active longitudes.