Only four tektite strewn fields have been produced during the past 40 m.y. even though at least 60 impact craters have been produced during the same period in continental lithologies having tektite-like compositions. The apparent reason for this discrepancy is that the ejecta from most crater-forming impacts was not completely melted. Laboratory and modeling studies show that projectiles as different as anorthosite and metal (densities from 2.8 to 8 g cm-3) produce similar amounts of melt, and this conclusion probably holds for comet-like densities near 1.5 g cm-3. Our interpretation of the limited evidence indicates that the influence of projectile trajectory and velocity on melting efficiency is minor. The key factor affecting melt production seems to be the nature of the target, particularly its porosity; the fraction of projectile kinetic energy converted to heat may be an order of magnitude higher in a highly porous target than in a void-free target. The grain size and water content of the target are also important. The ideal target is a porous, fine-grained sediment such as loess, particularly if it is also dry. We suggest that the rate of impact production of fully molten crater ejecta is proportional to the fraction of the continental surface having thick (>10 m) blankets of dry loess and that tektites are mainly produced during cold, dry climate periods when such deposits are an order of magnitude more common than at present. This model is best tested at the Australasian site, since detailed climate records cannot be obtained for the Czechoslovakian and North American sites and have not yet been obtained for the Ivory Coast site. The fossil O isotopic records in cores containing the Australasian microtektites show that these were deposited during a period of cold climate near glacial stage 20.2. The chemical homogeneity and the high 10Be contents of Australasian tektites are well explained by melting a loess target. ¿ American Geophysical Union 1993