Silage shrink (weight lost between ensiling and feedout) represent loss of nutrients to dairy producers, and the potential to degrade air quality if that loss is as volatile carbon compounds, or degrade water quality due to weepage to surface water and seepage to subsurface aquifers. No research has documented silage shrink in large commercial silage structures (silos) common in the SW US. ‘Shrink’ can be expressed as loss of wet weight (WW), oven dry weight (oDM) and oDM corrected for volatiles lost in the oven (vcoDM). Shrink losses, and the phase of the process where losses occurred, were measured using 7 corn silage silos (2 rollover, 1 bunker, 4 wedge) from the 2013 crop year on 4 dairy farms in 2 San Joaquin Valley areas, all covered within 48 h with an oxygen barrier inner film and black/white outer plastic weighted with tire chains. Total WW, oDM and vcoDM losses (not including wastage) calculated from weights of fresh chop delivered to the silo and silage placed in a feed mixer (n = 7) were 9.0 ± 1.69, 6.8 ± 1.82 and 2.8 ± 2.08%, suggesting that much of the WW shrink was water and much of the oDM shrink was volatiles driven off during oven drying. The largest part of shrink occurred in the silage mass (measured using in/out weights of 9–15 buried bags in each of 4 silos) before face exposure (WW, oDW and vcoDW losses from the mass were 3.9 ± 2.40, 7.2 ± 1.12 and 3.5 ± 1.27% respectively), with losses from the exposed face (measured as loss in core weight between freshly exposed faces and ~21 h exposed faces from 4 cores of 50 cm depth on 2 occasions in each silo), as well as between face removal and the mixer (measured between compositional changes between freshly exposed faces and silage placed in the mixer on 2 occasions in 4 silos), being negligible. Silo bulk density, face management, rate of face use and face orientation had no obvious effects on shrink. Real shrink losses (i.e., vcoDM) of well managed corn silages piles are much lower than has been generally assumed, the exposed face is a very small portion of those losses, and many of the proposed mitigations may not be effective in reducing shrink, possibly because it is quantitatively so small in large well managed silos.