Most theoretical models of dealer markets conclude that interdealer trade increases the efficiency of trading. Until recently, finance economists have not had access to the data necessary to assess the importance and consequence of interdealer trading in equity markets. This paper uses unique data from the London Stock Exchange to study interdealer trading. Besides documenting the extent of trading and the institutions that facilitate trade, we demonstrate that interdealer trding primarily corrects extreme inventory imbalances. We also find that interdealer trades differ in the benefits they confer to trade counterparties. These differences help explain the timing of interdealer trades.As for specific findings, on average interdealer trading accounts for approximately 28 percent (by volume) of all FTSE-100 trades, while interdealer trades account for 15 percent of trades in all other equities. Strikingly, we find that when a dealer’s inventory exceeds what we define as a normal level, interdealer trading increases to 65 percent of all trdes for a sample of FTSE-100 equities and 55 percent for a sample of Non-FTSE equities. We also show that dueing inventory imbalances, broker-dealers are much more likely to use all types of interdealer trades. Market makers appear to favor direct public trades when spreads are narrow. They favor anonymous interdealer trades when spreads widen or stock is liquid. We find market maker’s inventory management policies also differ by whether they are managing inventories before or after large customer trades. When market makers anticipate customer trades, they liquidate them much more quickly than when they do not anticipate them. When we compare the costs and benefits of different types of interdealer trades, we find that the consumers of limit orders in anonymous IDB systems receive more price improvement that market makers who deal directly.