Gravitational microlensing is a robust tool to detect and directly measure the abundance and mass of any kind of compact object, either in our galaxy or in the extragalactic domain. Based on generic, broadly applicable arguments, it is concluded that the observed microlensing magnifications are too small and the microlensing events less frequent than the expectations for a significant population of compact objects (other than normal stars). The detection of chromatic effects of microlensing also does not support the presence of black holes. Detailed statistical studies of the observed microlensing magnifications and events frequency impose strict upper limits to the fraction of total mass of black holes ( \(\lesssim \) 1%) from \(10^{-7}M_\odot \) to indefinitely large masses. These results hold even when the black holes are distributed according to a mass spectrum or are forming clusters.