Twenty-five pure pollens plus several blends of pollen were fed as sole protein sources to honey bees, Apis mellifera L., and resultant survival of bees on these diets was measured. Average increase in mean life span of bees on a pollen diet versus sugar water controls was 19.5 d, with a range from 3.9 d less than the controls for Ambrosia to 40.6 d longer, for a five-pollen blend. Actual consumption of test pollen diets also varied dramatically among test pollens, with a mean consumption of 16.5 mg pollen per bee for the first 10 d and a range of 1.9—29.0 mg per bee. Pollens that induced decreased life span included Ambrosia, Uromyces (a rust spore), Typha , and Kallstroemia ; those that induced little increase in life span included Haplopappus, Baccharis , and Taraxacum; and those that induced the greatest increase in life span included Ephedra, Prosopis, Rubus, Populus , and two blends of pollen. Increased life span was not correlated with pollen grain size, grain spininess, or dispersal vector (biotic versus abiotic) and was weakly correlated with season of floral bloom (pollen produced in spring was better than that produced in fall). Pollen from the family Compositae was poorer than average pollen. Major factors affecting life span were amount of pollen consumed, protein concentration in the pollen, and, especially, total amount of pollen protein consumed. By combining information relating to consumption levels and protein intake, reasons why a pollen caused the observed life span could be determined. Reasons included presence or absence of attractants and phagostimulants/deterrents, presence of toxic compounds, and a poor nutrient balance or level.