Motor vehicle dealerships in the United States tend to hold inventories equivalent to around 65 days' worth of sales, a relatively high level that has been nearly unchanged for 50 years. Despite playing a prominent role in the volatility of U.S. business cycles, very little is known about why the auto industry targets inventory stocks at such a high level. We use a panel of inventory and sales data from 41 vehicle brands over 30 years and the solutions to two well-known inventory planning problems to show that vehicle inventories appear to be related to (1) the size of dealership franchise networks, which tend to be large; (2) product variety, which tends to be high; and (3) the volatility of new vehicle sales, which also tends to be high. We show that differences across brands in these variables explain a good bit of the cross-section dispersion in brand inventory-sales ratios. Offsetting changes in these factors over time also help explain why the industry's overall inventory-sales ratio has been quite flat for many decades. More recently, the net increase observed in the inventory-sales ratio in the past couple of years is in contrast to fit of the model, which might suggest that some of that increase could reverse in the coming years.