Molecular gas in other galaxies is generally studied by observations of CO emission; a conversion from CO integrated intensity to H₂ column density must be made. Modelling of the emission from an ensemble of molecular clouds shows that these conversion factors are sensitive to temperature, so that molecular gas masses in galaxies with high star formation rates have probably been overestimated. Conversely, models of molecular clouds in low metallicity systems (such as irregular galaxies) demonstrate that the use of CO as a tracer can severely underestimate the molecular gas abundance. The observed properties of dark clouds and high latitude clouds are consistent with clouds in equilibrium with an intercloud pressure of P/k ≈ 10⁴. Detailed comparison of the CO and 170μm emission from the disks of NGC 6946 and M51 shows that the far-infrared flux must arise from dust in molecular clouds, not atomic clouds; this emission may be powered by embedded young stars or by the interstellar radiation field. The interpretation of the ratio of infrared to CO luminosities as a star formation efficiency is of dubious validity. Modelling of the observed CO and far-infrared emission from a sample of galactic nuclei shows that roughly half of the CO flux is produced by very active star-forming clouds with warm CO. The constraints placed on star formation models by abundance gradients in galaxies suggests that radial gradients in star forming efficiency generally exist in galaxies. The actual distribution of molecular gas in galaxies may be closely tied to the radial mass distribution.