In this paper, we develop a simple model that shows that consumption of PDS food grains is significantly different between rich and poor households in states where the PDS functions relatively well; in places where the PDS is non-functional, the difference is not significant. Using household-level data from three recent thick rounds of the consumption expenditure survey (2004-2005, 2009-2010 and 2011-2012), we find evidence in support of the predictions from the model. This suggests that one way to make the PDS functional is to make it more accessible to poor and underprivileged households.