Estimates of levels of Nosema apis infection in honeybee colonies by three different sampling and examination combinations were correlated with honey yield. Samples of 60 bees were taken in early spring in Uppsala, Sweden, live bees being taken from the cluster and dead ones from the bottom board. Parallel samples of hive bees were examined individually and compositely. Dead bees were only examined compositely. The mean spore count/bee of each composite sample was estimated with a hemacytometer. The correlation of honey yield with Nosema level was much larger for samples of live bees irrespective of counting methods, than for composite samples of dead bees. The closest correlation with honey yield (-0'707) was found for number of honeybees infected. It is concluded that when colonies are being examined in spring to determine the effects of Nosema on honey yield, it is best to sample live bees and not dead ones and to examine the bees of the samples individually rather than compositely.