A history of the idea of transporting large icebergs to arid regions to provide a fresh-water source is presented and the problem is considered in four main parts: (1) Location of a supply of icebergs. Only in the Antarctic are supplies of large tabular icebergs available. Data on the size distribution of these icebergs are reviewed and it is concluded that icebergs of almost any desired size can readily be located. (2) Towing. Steady-state towing velocities of different sized icebergs are calculated based on estimates of the drag of the icebergs and the bollard pull of tugs. Because drag increases with velocity squared, large icebergs can only be towed at very slow velocities (<c. 0.5 m/s). However, tugs that can be built within the capabilities of current technology are capable of towing extremely large icebergs. (3) Melting in transit. Calculations of melting indicate that, although melting losses are significant and may be excessive for small icebergs, when large icebergs are towed, large amounts of ice are left when the iceberg arrives at its destination. Towing trajectories, travel times, and ice delivery rates are calculated for optimum routes between the Amery Ice Shelf and Western Australia (A–A) and the Ross Ice Shelf and the Atacama Desert (R–A). Forces included in these calculations are towing, air, water, gradient current and Coriolis. Transit times exceed 107 d (A–A) and 145 d (R–A) with over 50% of the initial ice delivered. (4) Economic feasibility. After total towing charges are paid, it is possible to deliver ice to Western Australia for 1.3 mills/m3 of water and to the Atacama Desert region for 1.9 mills/m3. These costs are appreciably less than the expected price of water delivered at these locations (8 mills/m3). The water delivered by the operation of one super-tug alone would irrigate 16 000 km2. Problems related to both iceberg transport and processing are reviewed and although substantial problems do exist, they appear to be within the capabilities of current technology.