We have used data taken as part of the Herschel infrared Galactic Plane survey (Hi-GAL) to study 3171 infrared dark cloud (IRDC) candidates that were identified in the mid-IR (8 mu m) by Spitzer (we refer to these as 'Spitzer-dark' regions). They all lie in the range l = 300-330 degrees and vertical bar b vertical bar <= 1 degrees. Of these, only 1205 were seen in emission in the far-IR (250-500 mu m) by Herschel (we call these 'Herschel-bright' clouds). It is predicted that a dense cloud will not only be seen in absorption in the mid-IR, but will also be seen in emission in the far-IR at the longest Herschel wavebands (250-500 mu m). If a region is dark at all wavelengths throughout the mid-IR and far-IR, then it is most likely to be simply a region of lower background IR emission (a 'hole in the sky'). Hence, it appears that previous surveys, based on Spitzer and other mid-IR data alone, may have overestimated the total IRDC population by a factor of similar to 2. This has implications for estimates of the star formation rate in IRDCs in the Galaxy. We studied the 1205 Herschel-bright IRDCs at 250 mu m and found that 972 of them had at least one clearly defined 250-mu m peak, indicating that they contained one or more dense cores. Of these, 653 (67 per cent) contained an 8-mu m point source somewhere within the cloud, 149 (15 per cent) contained a 24-mu m point source but no 8-mu m source and 170 (18 per cent) contained no 24- or 8-mu m point sources. We use these statistics to make inferences about the lifetimes of the various evolutionary stages of IRDCs.