Northern fowl mites were monitored on a caged-layer operation in southern California for 22 mo. Three experienced observers underestimated actual numbers of mites in the vent region approximately 80% of the time. Errors were higher for heavy infestations. Observer estimates were highly correlated with each other (r > 0.89, P < 0.01) and with mite numbers estimated by vent feather removal (r > 0.82, P < 0.01). Mites on hens varied between houses and over time. Molting consistently reduced mite numbers, but did not eliminate then in a flock. Long-term monitoring of individual sentinel hens demonstrated that some hens would support high numbers of mites for several months or more. Use of a new sequential hen sampling plan required approximately 1 min per hen, if mite numbers were estimated. At this site, treatment decisions often could be reached in < 20 min per house. Mite scores (index of estimated mites per hen) were well correlated with percentage of hens infested in both test houses. In a chronically infested house, prevalence of mites on eggs averaged 8.5%, with a range of 0-55%. Applications of tetrachlorvinphos-dichlorvos by the producer appeared to be based on mites on > about 20% of eggs. The chemical was marginal for controlling mites on hens (25% reduction in percentage of hens infested), but effectively reduced mites on eggs (95% fewer mites on eggs at 1 wk and 90% at 2 wk). When data were grouped by mite index score on hens, there was a strong relationship (r2 = 0.83, P < 0.01) between mite prevalence on eggs and the scores of the hens which laid them. Sampling 100 eggs evenly spaced in a house required < 7 min, and adult mites were easily seen. Sampling mites on eggs appears to be useful to localize at least high-level infestations, and egg-based sampling for mites merits further investigation.