The observed mass function of young star clusters I will review observational constraints on the mass function of young star clusters (the "Initial Cluster Mass Function"; ICMF). Direct fits to sufficiently large samples of young star clusters suggest that the ICMF is well approximated by a Schechter function with low-mass slope dN/dM ~ M^-2 and an exponential cut-off at a few 100,000 Msun in quiescent environments (spiral discs). In starburst environments the cut-off appears to be at significantly higher masses, at least several million Msun, similar to old globular cluster systems. A Schechter ICMF is also consistent with observations of luminosity functions of cluster populations in spiral galaxies, that typically are steeper than dN/dL ~ L^-2, and with the observed correlation between age and luminosity of the brightest clusters in galaxies. The cut-off observed in the ICMF may be related to the upper mass limit of Giant Molecular Clouds which occurs at several million Msun, consistent with a low overall cluster formation efficiency within GMCs in spiral discs. Based on comparison with other spiral galaxies, the Milky Way should host about 20 clusters with masses > 10^5 Msun and ages < 200 Myr.