Horn flies, Haematobia irritans (L.), were photographed on cattle and then immediately estimated visually in the field on pastured beef cattle in southern California. Horn flies were counted in the pictures later on a computer screen. For 479 counts on individual cattle, the concordance correlation coefficient between the visual and photo-based assessments was 0.790, but was better for the higher half of the counts (0.732) than for the lower half of the counts (0.217). Major axis regression indicated that visual estimates were consistently higher than the number of flies counted in the pictures across the entire density range. Based on average raw means, the visual estimates averaged 21% higher than the photo counts. Visual estimates included flies on the belly and lower legs that could not be seen in a photo, and lower densities may have been more susceptible to such error. Where flies can be seen well, e.g. for very tame animals, the visual estimates were much faster and more costefficient and were sufficient to track relative horn fly abundance changes.