Measurements from space of the Earth's ultraviolet albedo are affected by scattering of sunlight from polar mesospheric clouds (PMC). We have examined 8 years of solar backscattered ultraviolet (SBUV) albedo data and find evidence of an annual occurrence of PMC in the summertime polar cap regions of both hemispheres. We have devised a cloud selection algorithm based upon the expected spectral properties of PMC scattering and show that the albedo residuals selected by this algorithm, at least above some threshold brightness, possess the basic properties of PMC with regard to their brightness levels, their seasonal characteristics, and latitude variation. Because of the uniform sampling of the polar caps, particularly in the southern hemisphere, it is possible to examine year-to-year variations in PMC activity. We find that a significant trend is present in the PMC occurrence frequency values over the period 1978 to 1986. The increase (up to a factor of 10 at some brightness levels) in the occurrence frequency from solar maximum to solar minimum conditions indicates an anticorrelation with solar activity, an effect that also appears to be present in noctilucent cloud sightings over the past several decades. Garcia  suggested that PMC should be modulated by changes in the solar Lyman-alpha (121.6 nm) flux through its strong photodissociation control of upper mesospheric water vapor. Solar Lyman-alpha underwent a significant (50%) variation during this period. Our results confirm this expectation, and thus indirectly support the current theories of PMC formation. However, other influences such as long-term changes of mesopause temperature or dynamics cannot be ruled out. We have identified two kinds of hemispherical asymmetries: the first, that PMC in the northern hemisphere is significantly brighter than in the south is consistent with previous results derived from Solar Mesosphere Explorer data; and the second, that the solar cycle response in the south is more pronounced than in the north. We show that if PMC scattering is not taken into account in the SBUV ozone retrieval algorithm, systematic errors of order 10% can occur in derived ozone concentration in the 40–50 km region of the summer polar cap.