We explore possible systematic errors in the mass measurements of stellar mass black holes. We find that significant errors can arise from the assumption of zero or constant emission from the accretion flow, which is commonly used when determining orbital inclination by modelling ellipsoidal variations. For A0620-00, the system with the best available data, we show that typical data sets and analysis procedures can lead to systematic underestimates of the inclination by ten degrees or more. A careful examination of the available data for the 15 other X-ray transients with low-mass donors suggests that this effect may significantly reduce the black hole mass estimates in several other cases, most notably that of GRO J0422+32. With these revisions, our analysis of the black hole mass distribution in soft X-ray transients does not suggest any "mass gap" between the low end of the distribution and the maximum theoretical neutron star mass, as has been identified in previous studies. Nevertheless, we find that the mass distribution retains other previously identified characteristics, namely a peak around 8M ⊙ , a paucity of sources with masses below 5M ⊙ , and a sharp drop-off above 10M ⊙ .