Water markets provide one of the most promising institutional mechanisms for increasing access to irrigation from groundwater, particularly for tenants and small farmers. While water markets are found in all provinces of Pakistan, they are most prevalent in canal irrigated areas of Punjab and in NWFP. This study reviews the emerging literature on water markets and uses farm-level survey data to examine the performance of groundwater markets, with particular emphasis on Faisalabad District in Punjab and Dir District in NWFP. Findings indicate that, while large landowners are more likely to own tubewells and pumps, smaller landowners and tenants are more likely to rely on purchases from other farmers' tubewells for access to groundwater. The distance over which water can be transported provides a limitation to water market sales, but lined watercourses increase the distance over which tubewell water can be sold. Contractual arrangements for water in the IFPRI study areas of Faisalabad and Dir districts include hourly charges, buyer providing the fuel plus a fee for wear and tear, and sharecropping for water. While all types of irrigation--canal, purchased groundwater, and own tubewell water--are shown to increase yields of wheat, groundwater has a higher impact than canal water, and water from own tubewells, which provides farmers with the greatest degree of control, has a greater effect on yields than purchased groundwater. Unreliability of access to purchased tubewell water was a problem for over half of water buyers in the study areas. This analysis indicates that purchasers are more likely to have unreliable access to groundwater if they buy water from small-capacity, electric-powered tubewells, if they are young and own little or no land. Policy measures to improve access to and reliability of groundwater through water markets include increasing the density of tubewells, especially by assisting small farmers to purchase tubewells; lining water delivery channels; and providing more reliable electrical power supply to rural areas. Further research is needed on how water markets work in less favorable environments, such as those with salinity, waterlogging, or falling water tables, and to identify policy interventions that are appropriate under each set of circumstances.