One of the great challenges in predicting the rate and geographical pattern of climate change is to faithfully represent the feedback effects of various cloud types that arise via different mechanisms in different parts of the atmosphere. Cirrus clouds are a particularly uncertain component of general circulation model (GCM) simulations of long-term climate change for a variety of reasons, as detailed below. First, cirrus encompass a wide range of optical thicknesses and altitudes. At one extreme are the thin tropopause cirrus that barely affect the short-wave albedo while radiating to space at very cold temperatures, producing a net positive effect on the planetary radiation balance and causing local upper troposphere warming, thus stabilizing the lapse rate. At the other extreme are thick cumulus anvil cirrus whose bases descend to the freezing level; these clouds produce significant but opposing short-wave and long-wave effects on the planetary energy balance while cooling the surface via their reflection of sunlight. In fact, satellite climatologies show a continuum of optical thicknesses between these two extremes (Rossow and Schiffer 1991). In a climate change, the net effect of cirrus might either be a positive or a negative feedback, depending on the sign and magnitude of the cloud cover change in each cloud-type category and the direction and extent of changes in their optical properties (see Stephens et al. 1990). Second, the dynamic processes that create cirrus are poorly resolved and different in different parts of the globe. In the tropics, small-scale convective transport of water from the planetary boundary layer to the upper troposphere is the immediate source of a significant fraction of the condensate in mesoscale cirrus anvils (see Gamache and Houze 1983), and ultimately the source of much of the water vapor that condenses out in large-scale uplift to form thinner cirrus. However, many observed thin cirrus cannot directly be identified with a convective source, suggesting that in situ upper troposphere dynamics and regeneration processes within cirrus (see Starr and Cox 1985) are important. In mid-latitudes, although summertime continental convection is a source of cirrus, in general cirrus is associated with mesoscale frontal circulations in synoptic-scale baroclinic waves and jet streaks (see Starr and Wylie 1990; Mace et al. 1995).