We summarize indirect empirical arguments used for estimating black hole (BH) masses in ultraluminous X-ray sources (ULXs). The interpretation of the X-ray data is still too model-dependent to provide tight constraints, but masses <~ 100 Msun seem the most likely. It is getting clearer that ULXs do not show the same evolutionary sequence between canonical spectral states as stellar-mass BHs, nor the same timescale for state transitions. Most ULX spectra are consistent either with a power-law-dominated state (apparently identical to the canonical low/hard state), or with a very high state (or slim-disk state). Despite often showing luminosity variability, there is little evidence of ULXs settling into a canonical high/soft state, dominated by a standard disk (disk-blackbody spectrum). It is possible that the mass accretion rate (but not necessarily the luminosity) is always higher than Eddington; but there may be additional physical differences between stellar-mass BHs and ULXs, which disfavour transitions to the standard-disk, radio-quiet state in the latter class. We speculate that the hard state in ULXs is associated with jet or magnetic processes rather than an ADAF, can persist up to accretion rates ~ Eddington, and can lead directly to the very high state. Comment: 8 pages; to appear in the proceedings of the conference "Observational Evidence of Black Holes", Kolkata, February 2008