Shallow clouds covering vast areas of the world's middle- and high-latitude oceans play a key role in dampening the global temperature rise associated with CO.sub.2 . These clouds, which contain both ice and supercooled water, respond to a warming world by transitioning to a state with more liquid water and a greater albedo, resulting in a negative "cloud-phase" climate feedback component. Here we argue that the magnitude of the negative cloud-phase feedback component depends on the amount and nature of the small fraction of aerosol particles that can nucleate ice crystals. We propose that a concerted research effort is required to reduce substantial uncertainties related to the poorly understood sources, concentration, seasonal cycles and nature of these ice-nucleating particles (INPs) and their rudimentary treatment in climate models. The topic is important because many climate models may have overestimated the magnitude of the cloud-phase feedback, and those with better representation of shallow oceanic clouds predict a substantially larger climate warming. We make the case that understanding the present-day INP population in shallow clouds in the cold sector of cyclone systems is particularly critical for defining present-day cloud phase and therefore how the clouds respond to warming. We also need to develop a predictive capability for future INP emissions and sinks in a warmer world with less ice and snow and potentially stronger INP sources.